Axsys Signals User Guide

A simple guide for everyone who uses Axsys Signals

0.Document control

0.1Changelog

0.2What this document is

This guide shows you how to use Axsys Signals. It covers signing in, reading your queue, opening an account, rating what you see, and what happens to what you type. You do not need to know anything about computers. You need a web browser and your email.

If you set up or run the server, this is not the right guide for you. Ask for the operations guide instead.

When something is not built yet, this guide says so. When a limit is there on purpose, it says that too. We would rather you know than guess.

This guide describes Axsys Signals release 0.5.1, as of 2026-09-28. The guide has a version of its own, 0.10, printed at the top and the foot of this page. The two count different things: 0.5.1 is the product you are signed in to, 0.10 is this edition of the writing about it.

1.Welcome

Overnight the desk reads public sources against who you sell to. Filings, news, job posts and more. It looks for events that tend to happen before a company buys. Then it ranks your accounts by how well they fit you and how soon a buying window looks likely to open.

You open the queue. You see who moved and why. Every score comes with its reasons and its evidence. You act on the hottest accounts first, and you leave the rest to be watched.

You do not need any computer skills for this. If you can read an email and click a link, you already know everything you need.

2.Signing in

2.1What you need

You need two things:

• An invitation from your workspace admin, or an Axsys account that has already joined a workspace.
• Your email address and your Axsys password.

The address is https://signals.srv1195619.hstgr.cloud. Save it in your browser. It works from any computer with the internet. You do not need to be in an office or on a company network.

Who needs an invitation, and who can just sign in:

• You have never used Axsys Signals. There are two ways in. If your company already has a workspace, your admin invites you from the Team page and you look for an email with the subject "You are invited to (your workspace) on Axsys Signals". If nobody has, you can start your own workspace at https://signals.srv1195619.hstgr.cloud/signup with your work email. Free mailboxes such as gmail are refused, on purpose: this is a product for a company, and the workspace is the company's. If a work email is a problem for you, ask for access instead (section 2.7).
• You have already joined a workspace. Go to the address and sign in with your email and your Axsys password. Nothing else is needed.
• You have an Axsys account from another Axsys product, like the CRM, but you have never joined a Signals workspace. You still need an invitation. Your account alone is not enough. Without an invitation you will see "No workspace yet". Section 2.5 explains what to do when the invitation arrives.

2.2Signing in

1. Go to https://signals.srv1195619.hstgr.cloud. You land on the sign in page.
2. Type your email and your Axsys password.
3. Click Sign in.

You land on Today, your queue. If you belong to more than one workspace, you first see a page called "Choose a workspace". Click the one you want.

Under the sign in form are three lines. "Forgot your password?" is section 2.4. "New here? Start your own workspace, or ask your workspace admin to invite you. Already have an Axsys account? Sign in with that email and password." means exactly what it says. "Stuck? Help with signing in" opens the troubleshooting page at https://signals.srv1195619.hstgr.cloud/help/troubleshooting, which needs no sign in.

2.3If it will not let you in

What you see, what it means, and what to do.

2.4Forgot password

1. On the sign in page click Forgot your password?
2. Type your email and click Send reset link.
3. You see "Check your email". This page always says the same thing, even when the email is not known to us. That is on purpose, so nobody can use it to find out who has an account.
4. Open the email called "Reset your Axsys Signals password" and click Set a new password. The link works for one hour.
5. Choose a password of at least 10 characters and click Set password.

You are signed in straight away and land on Today.

Two things to know:

• This also changes your CRM password, if you have one. Axsys Signals and the Axsys CRM share one account per email. One password, two doors. This is on purpose, not a mistake.
• Setting a new password signs you out everywhere else. Sign back in on your other devices with the new one.

If the email does not arrive within a few minutes, check your spam folder. We send at most three reset emails per address per hour; extra requests are dropped quietly, so wait an hour and try once. If you see "Link expired", the link is older than an hour, was already used, or your password changed after it was sent. Ask for a new one.

2.5Accepting an invitation when you already have an account

If you already have an Axsys account with the invited email, the invitation page says "Join (workspace)" and asks for Your Axsys password. Type your existing password and click Join workspace. There is no Microsoft sign in here. The Axsys account is its own.

If you never set a password (for example, you only ever used Microsoft sign in on the CRM), do this:

1. Go to https://signals.srv1195619.hstgr.cloud/forgot and reset your password first (section 2.4).
2. Open the invitation link again.
3. Type the new password and click Join workspace.

If you open the invitation before doing that, the page offers the "create your account" form. When you submit it, a page called "This email has no password yet" appears with a Reset password link. It means exactly that. Follow the three steps above.

If you are brand new, the page asks for your name and a password of at least 10 characters. Click Create account and join.

If you see "Invitation not valid", the link has expired (links last 14 days), it was withdrawn, it was already used, or your admin sent you a newer one (only the newest link works). Ask your admin for a new invitation.

2.6Signing out and dark mode

Both live at the bottom of the left rail, under your name and role. Click Sign out to leave. Click Dark mode to switch the colors, and Light mode to switch back. Your choice is remembered in that browser only.

2.7Ask for access

Most people should start their own workspace at https://signals.srv1195619.hstgr.cloud/signup. Ask for access is for the cases that does not cover: you would rather talk to a person first, or the only address you have is a personal one. Go to https://signals.srv1195619.hstgr.cloud/request. The page is called Ask for access. It asks for four things: your name, your work email, your company, and one line on what you sell. Press Send.

You then see "Thanks. We will write back within two working days." That is the whole promise: a person reads it and writes back. Nothing is created for you on the spot, and no password is set. If we say yes, you get a normal invitation email (section 2.5).

Two small rules. You can send three requests an hour from one network, and three a day for one email address. A free email address, like gmail, is accepted, but we notice it, so a work address gets a faster answer.

3.Finding your way around

3.1The left rail

After you sign in, the dark blue column on the left is the rail. It has nine items:

• Today. Your ranked queue. The screen you land on. Section 4.
• Signal desk. Where you mark a lead Good or Bad with a reason, and stamp each kind of signal 1 to 5. Section 6.
• Library. The catalog of every event we can watch for. Section 7.
• ICP. Who you sell to, in one statement. Section 8.
• Your signals. What the desk watches for you, by category, with a switch for each kind of event. Section 9.
• Sources. The places we read to find evidence, one row each, with a switch for each channel. Section 10.
• Health. Did each collector run, did it find anything for you, and why not. Section 11.
• Team. Who is in the workspace, inviting people, and the workspace settings. Section 13.
• Guide. This guide.

The item you are on is marked with a thin white bar. Discovery and Workshop are not in the rail. They are the optional deep discovery, section 14, and you reach them by their addresses. If your admin keeps the ICP private, the ICP item is not in your rail either (section 8.4).

3.2The workspace card

At the bottom of the rail is a card labelled WORKSPACE. It shows the name of your workspace and one of:

• Trial, N days left. Your workspace is on a trial. The number counts down.
• Active. Your workspace is live.
• Read only. The trial ended. You can still look at everything. You cannot rate, change the set or invite anyone. Every save shows "This workspace is read-only. Contact Axsys to continue." Three things do still work, on purpose: you can save a lead as a CSV or JSON file, you can use Report a problem to tell us something is wrong, and the first login tour still closes for good once you have seen it. Sending a lead to your CRM is not one of them. Write to the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai to extend it.

Under the card: your initials, your name, your role, Sign out and Dark mode.

3.3The building blocks

Every screen is made of the same few pieces. Once you know them you can read any page.

• Tile. A big number with a small label above it, in a band across the top of a page. Four tiles sit side by side. They are counts, nothing else.
• Chip. A small square stamp in capital letters, like TIER 1 or IN YOUR SET. Some chips are buttons, like the filters at the top of the queue. A dark filled chip is the one that is switched on.
• Ruled row. One line in a list, with a thin rule under it. Rows in the queue can be clicked. A thicker double rule closes a list; that is where the list ends.
• Drawer. A panel that slides in from the right when you click a row or a card. Close it with the Close button, the Escape key, or by clicking the grey area beside it.
• Glyph. A small mark before a status word: a triangle, a diamond or a dot. Section 4.2 explains them.
• Notice band. A pale amber band at the top of a page. It carries a note from the team, or the sentence "Live data is delayed". Nothing is lost when you see it.
• The two buttons at the top right. Take the tour (section 3.5) and Help (section 3.6) sit at the top right of every signed in page.

3.4If a button is not there

If you expect a button and it is not there, it is almost always your role, not a mistake. Viewers see everything but change nothing. Contributors can rate and shape the set. Admins can also invite and manage people. Section 16 has the table. Ask your admin if you are not sure.

3.5The tour

The first time you open Today, a short tour starts by itself. A spotlight lights up one part of the screen at a time, with a small box that explains it. The box has Back, Next and Skip, and a count like "Step 3 of 5". It takes about a minute.

Every step is on Today, so the tour never carries you off to another page. There are five steps: your list, opening a row, the proof behind the score, getting the lead out, and the rest of the app. If your workspace has no accounts yet there are three steps instead: why the list is empty, what to read while you wait, and where to find the tour again.

You can start it as often as you like. Take the tour sits at the top right of every signed in page, and the same words are the third entry in the Help menu beside it (section 3.6). Press it anywhere else in the app and Today opens with the tour running. The button is filled in dark until you have seen the tour, then it goes quiet, so it never nags you. The one page without it is the ICP setup screen, because there is no queue to walk through until a profile is approved; Help is still there.

It starts by itself only once. When you press Done on the last step, or Skip at any point, it is marked as seen for you. If your workspace is still waiting for its admin to approve the ICP (section 8.3), the tour waits too, because there is no queue to show yet.

3.6The help button

Help is a button at the top right of every signed in page, beside Take the tour. The round ? that used to float in the bottom right corner is gone; it covered a switch on Sources, Remove on Team, and a filter chip on a phone. Click Help and a small menu opens with five choices:

• Ask Signals. Opens a panel on the right where you type a question about your own workspace and read the answer. Section 3.7.
• Guide for this page. Opens this guide at the chapter for the screen you are on. On Today it opens chapter 4, on Sources chapter 10, and so on.
• Take the tour. Runs the tour again (section 3.5).
• Troubleshooting. Opens the troubleshooting page, the list of every message the app can show and what to do.
• Report a problem. Opens a small form: What is it about (a few words), What happened (up to 2,000 characters), and the page you are on, filled in for you. Press Send. You see "Sent. Reviewed on Thursdays."

That last line is the promise. Reports go to the Axsys team's issue list and are reviewed on Thursdays; they are not read by a machine and nothing changes by itself. Say what you did, what you expected and what you saw. You can send ten reports an hour. Any member can report.

3.7Ask Signals

Ask Signals is the first choice in the Help menu (section 3.6). Pick it and a panel slides in from the right, the same shape as the account drawer. The first line in it says: "Answers come from this workspace's own data. Nothing you ask changes a score." Type a question in plain words and send it. The answer comes back as a sentence or two, in plain words, and you can ask a follow up in the same panel. The conversation stays in that browser tab for as long as it is open. Close the tab and it is gone; the panel starts clean next time.

It answers about your own workspace, and only that. Ask it about:

• Your queue. How many accounts are on it, how many sit in each tier, what the status words mean, and why the list is the length it is.
• An account and its score. Why an account sits where it does, what each part of the score earned, its biggest gap, what would move it up, and the newest events on its timeline. Name the company and it finds the account; if the name is not in your workspace, it says so.
• Results. What came in for an account, in a family or of a kind, in the last so many days, and counts by channel, family, account or kind.
• Sources and switches. Which channels are on or off, which kinds of event are switched off on Your signals, and the state of the sources you added yourself.
• The ICP. Who you sell to, the filters in plain words, and who is on the list. If your admin keeps the ICP private (section 8.4), it says "Your admin keeps the ICP private." and stops there.
• Whether a collector ran. The same rows as the Health page (chapter 11). That one can take up to twenty seconds, because it asks the collectors directly.
• The Library and this guide. What a kind of signal means, its rating and lead time band where the library has them, and where in this guide or the manual a thing is explained.

Three things it will not do, on purpose:

• It changes nothing. It does not rate a lead, flip a switch, send a lead to your CRM or touch a score. It only reads. There is no button in the panel that saves anything.
• It cannot see another workspace. Every answer is built from your workspace's own data, the same data every page you open is built from. There is no way to point it at anyone else's.
• It does not invent a number. Every count and every score in an answer comes from the data, added up by the app, not guessed by the model. When the data cannot answer a question, it says so rather than filling the gap.

There are two limits: twenty questions an hour per person, and sixty a day for the whole workspace. When you reach the first, the panel says "Slow down a little." When the workspace reaches the second, it says "This workspace has used its questions for today." Nothing is lost; the pages themselves keep working, and everything the panel reads is also on Today, the brief, Sources, Your signals, Health and the Library.

Any member can ask, viewers included. It sees only what you can already see. A workspace that is read only can still ask.

Axsys logs each question and its answer, the same way it keeps your ratings, and reads them in the weekly review. That is how we learn which questions people bring and where the answers fall short. Do not type a password or anything private into the panel.

If the panel says "Ask Signals is not available right now.", the model that writes the answers could not be reached. The rest of the app is not affected; try again in a few minutes. If the choice is not in the Help menu at all, it is switched off on the server. The troubleshooting page has a row for every message the panel can show.

4.Today, your queue

4.1What Today is for

Today is the first screen you see. It is your ranked list of accounts, best first. Use it to see who moved this week, why, and what to do about it. Most days you only need this screen and the brief behind it.

4.2What each thing on the Today screen means

Across the top:

• The date line under the word Today shows the day the queue was scored.
• Search accounts and signals. Type any word. The list narrows to accounts whose name, industry or signal text contains it.
• Invite teammates. Admins only. It takes you to the Team page. Nobody else sees this button at all.

The four tiles:

The filter chips, under the tiles. One is on at a time:

To the right of the chips: "N of M accounts, ranked by fit + intent". N is what the chip shows, M is the whole queue.

Each row in the queue, left to right:

The status glyphs:

Surging, Heating and Monitoring are the same scale as the tiers, with three words instead of four numbers. Surging is Tier 1. Heating is Tier 2. Monitoring is Tier 3 and Tier 4. Two names for one thing; you are not missing anything.

If no row matches the chip you are on, you see "Nothing matches. Try another filter."

If your workspace has no accounts at all yet, it says something different, because no filter can help: "No accounts yet. We look overnight, so your first list lands in the morning. Nothing is broken." Admins get a link to the ICP page under it, so they can check who the queue is looking for.

4.3The drawer

Click any row. A drawer slides in from the right with the account name at the top and a Close button. It has five panels.

1. The header chips. Tier and its status word (Pursue now, Develop, Monitor or Low priority), the glyph, the score, the timing window (Next 30 days, Next 90 days, Next 120 days, or Beyond 120 days), and the industry.
2. Recommended action. One sentence that depends on the tier. Tier 1: reach out this week, and open with the most recent event, which is named. Tier 2: develop the relationship; timing is not proven yet. Tier 3 and 4: monitor, light touch only.
3. Why this score. The card that says where this account sits and why, part by part. The table under this list reads it top to bottom.
4. Account facts. The segment (SMB, Mid-market or Enterprise), the head office and the revenue range, then Evidence: how many observations are listed below (at most 8), and the confidence label.
5. Signal timeline. One line per observation, newest first, at most 8 (older ones are kept in the platform but not listed here): the date, what we found, a short excerpt when we have one, then the evidence family, the source and the confidence, and a points mark like +8. At the end of the line is Source, a link to the page we read, when we have one. When we do not, the line says "no link on file". The points show how much that kind of signal usually matters. They are not added up into the score. The score is fit plus timing, nothing else.

The Why this score card is the same card in the drawer and on the brief. Its head says "Fit, evidence and timing, out of 85." (or out of 100 for a workspace with no size band). Under that it has four parts:

The line of detail under each row is the useful part. The size row names the headcount and says whether it is inside your band. The evidence row names the areas we have found something in. The profile row says whether we have the industry and whether we have checked the website. The timing row says where the estimate came from, or that there is none yet.

A real account, from the day this build went live. N-able scores 75 out of 85 and sits in Pursue now. Its card takes full marks on how much we know about it, the spread of evidence, the strength of its best signal, how much evidence there is and how recent the evidence is, and it loses its ten points across the two parts that are short: how many sources confirm it, and when it is likely to buy. Because it is already Tier 1 there is no line about moving up.

An account further down reads the same way with more losses: the sentence names each part it is short on with the points, the rows show where, and the foot says how many points would carry it into the band above and which part has the most room. On Today, that account's row prints the biggest gap in a few words, for example "Biggest gap: one source only" or "Biggest gap: their evidence is a few months old". Section 5.3 says what you can do about a card that reads like this one.

One thing worth getting straight, because it catches people out. The numbers in this card are the score. The small +8 or +4 marks beside each result in the timeline below are a different number altogether: they say how heavy that kind of signal is in our catalog, and they are never added up. Add the marks in the timeline and you will not get the score, and you are not meant to.

When we cannot put a plain sentence together for an account, the card says so, in these words: "We do not have a plain summary for this account yet. These are the parts of the score, as the engine records them." The rows are still there with the engine's own labels. We would rather tell you we cannot explain it than make something up.

The tier ranges are Tier 1 80 to 100, Tier 2 65 to 79, Tier 3 50 to 64, Tier 4 under 50.

Two confidence labels appear. Strongly Supported means this kind of signal usually shows up in more than one independent stream at once. Inferred means one stream suggests it but does not prove it. Today the label comes from the signal's definition in our catalog, not from a check of that one result. Treat it as a hint about the kind of signal, not proof about this account.

Under the panels are two buttons: Open the full brief (section 5) and Rate these signals on the desk (section 6). Next to them are three more for getting the lead out: Download CSV, Download JSON and Send to CRM. Chapter 12 explains all three. Send to CRM is grey until it can be used, and the reason is printed as a line under it, not just as a tooltip. An admin with no webhook saved yet reads "No webhook set. Set your CRM webhook on the Team page." with a link to Team. Everyone else reads "No webhook set. An admin sets it on the Team page." A viewer reads "Only an admin or a contributor can send." Viewers do see the button; it is grey and it does nothing.

4.4What to click next on Today

• Start with This week's movers. Click the top row. Read the recommended action. Click Open the full brief if you want the whole picture.
• Switch to Pursue now when you only want Tier 1.
• Use Thin evidence to find accounts you should not act on yet.

4.5If Today gets stuck

• The queue looks the same as yesterday. The queue is rebuilt from our data about every 60 minutes. New signals land in it once a night: at 06:30 UTC the app carries in what the collectors found and scores every company again, so the first view of your working day is a fresh one. A morning with no change means the night was quiet for your accounts, not that nothing ran. The Health page (chapter 11) says which collectors ran and what they found.
• Only one or two accounts are Tier 1. Tier 1 needs a high score, and a high score needs evidence from several different areas plus timing points. Most accounts do not have that yet. Open any account that is not Tier 1 and the Why this score card says which parts it is short on and by how many points (section 4.3); section 5.3 says what you can do about each one. See also the first question in section 17.
• A pale band says "Live data is delayed". You are looking at the last good copy. Nothing is lost. Refresh in a minute or two.
• The brief shows "No such account in the current view". The account dropped out of the top 40 since the page loaded. Go back to Today and look again.

5.The account brief

5.1What the brief is for

The brief is the full page for one account. It shows everything the drawer shows, with more room, plus the mix of evidence. Open it from the drawer with Open the full brief.

5.2What each thing on the brief means

At the top: a Today link to go back, the account name, and a line with the industry, the segment and the head office. On the right, the Tier chip with its status word, the glyph word and the score.

Then the Recommended action band, the same sentence as in the drawer, with a Rate on the desk button.

Under the band sits one ruled bar with two halves. On the left, three buttons for getting the lead out: Download CSV, Download JSON and Send to CRM (chapter 12). On the right, your call on the lead: Good lead and Bad lead, your verdict once you have given one, and the team tally, like "Team: 2 good, 1 bad". Section 6.2 explains the reasons. Viewers see the tally only.

Under that bar, across the whole width of the page, sits Why this score: the same card as in the Today drawer, with the same sentence, the same row for every part of the score, the same Score out of its maximum at the foot and the same line about what would move the account up. Section 4.3 reads it line by line, and section 5.3 says what you can do about it. On the brief it has the full width, so the long labels are not squeezed.

Below, two columns.

Left, Signal timeline. The heading says how many observations are in the window. Each line has the date, what we found, a short excerpt when we have one, the evidence family, the source and the confidence, the points mark, and at the end Source, a link to the page we read. When we have no link, the line says "no link on file" instead. The brief shows at most 8 observations per account, newest first. Older ones are kept in the platform but not shown here.

If you came here by clicking a result on the Signal desk, that result is highlighted in the timeline and the page scrolls to it, so you land on the line you clicked.

Right, two cards:

• Account facts. The window (Next 30 days and so on) and Revenue, which says "Not known" when we have no figure. That line was labelled Ownership for one release while it held the revenue range, which was simply the wrong word; it now says Revenue. We hold no reliable ownership record for most companies, so the brief shows none.
• Signal mix. One bar per evidence family, showing what share of this account's observations comes from that family. More families means broader evidence, which is one of the things that lifts fit.

Two real limits, stated plainly:

• Not every result has a link. Our collectors only started keeping the page address recently, so older results say "no link on file". They fill in over time as the collectors run. Where there is a link, the excerpt is the part of the page we matched, not the whole article.
• There are no people or contacts on the brief. It is about the company. Contacts are a later step.

5.3What you can do about a low score

The card tells you which part is short. Here is what each part really rests on, so you know what is worth waiting for and what is not.

• The spread of evidence. This one fills in on its own. We read our sources every night, and a new kind of proof about the company counts the day it turns up: an account with evidence in one area reaches two areas as soon as something different lands. There is no button that makes it happen faster. One thing is worth checking, though: if your workspace has switched off a kind of event (chapter 9) or a whole channel (chapter 10), those results are dropped before the score is worked out, so a switch of your own can be what is holding the number down.
• When they are likely to buy. This is the part most accounts lose the most on, and it needs a real event with a date on it. Until something dated turns up that our scorer can put a lead time to, there is no estimate, and the points are worked out from how strong the signals are instead. That path never reaches more than 24 of the 25 points, so an account with no timing estimate has to take full marks on everything else before it can reach Pursue now at all. Few do, which is why timing is the biggest gap on so many rows. Waiting for the event is the honest answer.
• How much we know about them. Two things: their industry, and whether we have checked their website. Both fill in as we collect. If you know the company and the card says we do not have its industry or have not checked its site, tell the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai. That is a hole in our records, not in the company.
• Their size. This is the company's headcount against the size band in your profile. The company will not change, so the only thing in your hands is the band, and only if the band is wrong. An admin can edit it on the ICP page (section 8.3). Widen a band because it is truly wrong, never to make one number look better; you will only fill your queue with accounts you do not want.

Two things we do not do, said plainly. There is no button that sends us off to research one account on demand. And rating a lead Good or Bad does not move the score: your ratings go to the Axsys team's weekly review (chapter 15), and nothing in the scoring changes by itself.

5.4What to click next on the brief

Read the recommended action. If you agree with the evidence, act on it: press Send to CRM, or download the lead as a file (chapter 12). Then say what you think with Good lead or Bad lead and a reason. If you want the wider context, go to the Library and open the signal's category.

5.5If the brief gets stuck

• "No such account in the current view." The account is not in your current queue. Go back to Today.
• The timeline says "No observations." The account is in the queue on its timing note alone. Treat it as thin.

6.Signal desk

6.1What the desk is for

The desk is where you tell us what you think. Two things happen here. You mark a lead Good or Bad, with a reason. And you stamp each kind of signal with a value from 1 to 5 and say why. All of it is saved with your name. Your team sees the tally and the team average, not the names.

Said plainly, and the desk says the same in its own footer: your marks and stamps are saved for the Axsys team's weekly review. Nothing in the scoring changes by itself. Section 15 explains the loop.

6.2Results to review

The first card lists the twelve newest results across your accounts. Each line shows what we found, a short excerpt when we have one, then the account, the source, the date, the evidence family, the points mark, and Source, a link to the page we read. When we have no link the line says "no link on file". Click anywhere on the line and the brief opens with that result highlighted (section 5.2).

On the right of each line are two buttons:

• Good lead. This company is worth a salesperson's time.
• Bad lead. This company is noise, or not a buyer for you.

Click one and a small picker opens under the line: "Good lead · why?" or "Bad lead · why?", a row of reason chips, and a box "Why? (240 characters)". You only ever see the reasons that suit the button you pressed. Good lead shows four chips: Good fit, Good timing, Strong evidence and Other. Bad lead shows six: Not a fit, Wrong size, Wrong industry, Old news, Already a customer and Other. Pick every reason that applies: one, two or all of them. A chip stays on until you press it again. You may add a sentence. Press Save. The line then shows "You: Good lead, good fit and good timing" (or whatever you picked) and the team tally, like "Team: 2 good, 1 bad". Save again with the other button to change your mind; one verdict per person per lead.

The nine reasons in all, and which button shows each one:

Two things to know. Your mark is on the lead, which is the company, not on one line. If two lines belong to the same company, both show the same mark. And, said plainly: "Saved for the Axsys team's weekly review. Nothing in the scoring changes by itself." A hundred Bad leads do not move a score. A person reads them each week and decides.

Admins and contributors can vote. Viewers see the tally and no buttons.

6.3The source library

The second card is a table of the kinds of signal that produced results in your window, grouped by evidence family. Columns:

The card header carries a small link, Manage on Your signals. It opens the page where your workspace switches a kind of event on or off (section 9). The places we read, the channels, are managed on the Sources page (section 10). A kind that was switched off usually does not appear in this table at all, because its results were dropped before the table was built. Where a row does carry an Off chip, shown dim, that kind is off for your workspace.

After you click a stamp, a box appears: "Why that value? (280 characters)". Type your reason. It saves by itself about a second after you stop typing, and a small "Note saved" appears. Under the stamps you see "Team avg X, N votes" or "No votes yet". Click the same stamp again to clear it.

Ratings are shared with your team as a team average and a count of how many rated. The desk never shows who rated or what they wrote. Each rating is still stored with your name, so the Axsys team can see who said what in the weekly review.

6.4What to click next on the desk

Work down the results list. Mark what you can, with a reason, and skip what you cannot judge. Click Source when you want to read the page yourself. Then stamp the two or three kinds of signal you feel strongly about, and say why in a sentence. Ten honest notes are worth more than fifty quick clicks.

6.5If the desk gets stuck

• "Slow down a little." You have saved more than 240 marks or stamps in an hour. Wait a bit.
• "Pick at least one reason." You pressed Save with no reason chip on. Pick one or more.
• "With this many reasons the line is 12 characters too long." Your reasons and your sentence share 280 characters. Shorten the sentence a little, or pick fewer reasons.
• "Only an admin or a contributor can vote." You are a viewer. You can still see the tally.
• "This workspace is read-only." Your trial ended. Section 3.2.
• "No results in the window yet." Nothing has been observed on your accounts recently. Check Today for the wider queue.
• A line says "no link on file". We have no page address for that result. Not an error; section 5.2 says why.
• The Status column is grey with an old date. That kind of signal has found nothing new for you lately. To see whether its collector ran, open Health (chapter 11).

7.Library

7.1What the Library is for

The Library is the catalog of every event we know how to watch for. It is organised as categories, the signals inside each category, and the data sources that feed each signal. You use it to understand what a signal on the queue actually means, and to decide which signals belong in your set.

7.2What each thing on the Library screen means

At the top, the line under the title says which industry the grades are for (for example "Graded for technology"), or "Set your workspace industry to see grades" when no industry is set. The industry comes from the first industry in your approved ICP, and from the industry Axsys set on your workspace only when your ICP names none. The search box, Search signals, categories, sources, narrows the cards to signals whose name or text contains your word.

The four tiles:

On the left, a rail lists All categories and then each category with its signal count. Click one to see only its signals.

The grid on the right shows cards:

• With All categories picked and no search, you see one card per category: a CATEGORY chip, a grade (or "Not graded yet"), the name, a short cause and effect summary, and how many signals and sources it holds.
• With a category picked, or a search typed, you see one card per signal: a SIGNAL chip, a state chip (In your set, Added by you, Modified, or Excluded), a grade, the name, the summary, and chips for the chain grade and evidence tier. "Below the bar" marks a signal that did not pass our own validation and is shown for completeness.

A grade is a number like 7.1. Grades are per industry. A 7.1 in one industry is not a 7.1 in another. The grade line on a card says what it is based on. Today that is a synthetic backtest, not your own results. It gets better as real results come in.

7.3The three card types and the chain

Click any card. A drawer opens. There are three kinds of card, all built the same way, so once you can read one you can read all three.

• Category card. The domain it belongs to, the grade, the summary, then two lists of chips: Signals in this category and Data sources that feed it. Click any chip to open that card. A Back button takes you up a level.
• Signal card. Its category (click it to go up), its state in your set, the grade, the summary, and then The causal chain: three boxes labelled TRIGGER, MECHANISM and PREDICTABLE NEXT STEP. Read them left to right: this happens, which causes this, which usually leads to that. A "Weakest link" line says where the chain is most likely to break. Under it, Where we watch for it: a table with the source, the threshold that must be crossed, which problem it feeds, lead time, shelf life, and whether the signal forces action or only suggests it. Then chips for chain grade, evidence tier, discoverability, business function, buyer, and a validation score out of 5.
• Data source card. Its kind (filing, news, job postings, web, registry or api), how often it is read, the grade, the summary, a cost note and a link to where it lives, and Signals this source feeds.

Lead time is how long before a buying window this signal usually shows. Shelf life is how long it stays useful. Many rows have these blank today, because per signal values are not loaded yet. A source shown as "Unassigned" means the chain row has no data source linked yet.

7.4Keep in set and Exclude

At the bottom of a signal card is Your set with two buttons: Keep in set and Exclude. Admins and contributors can click them. Viewers see the buttons greyed out with the note "Admins and contributors can change the set".

Exclude marks the signal as not yours. The card shows EXCLUDED and the In your set tile goes down by one. Keep in set puts it back. Every change is saved with your name.

A limit, stated plainly: today your set changes what the Library shows and counts. It does not change which signals reach your queue. The Library's list and the desk's catalog are two different lists, and the link between them is not built yet. To switch a kind of event off in your queue, use Your signals (section 9). The card says so under the two buttons: "Keep and Exclude shape your library set. To switch a kind of event off in your queue, use Your signals."

Contributors and admins can also add a signal of their own to a category. There is no button for it on the screen yet; ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai and we add it for you.

7.5If the Library gets stuck

• "The library is empty for this workspace." The catalog has not been loaded for you. Tell the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai.
• "Set your workspace industry to see grades." Neither your ICP nor your workspace record names an industry. An admin can add one to the ICP (section 8.3), or ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai to set it on the workspace.
• "Not available right now." The data provider is slow. Wait a minute and reload.

8.ICP, who you sell to

8.1What the ICP page is for

ICP stands for ideal client profile: the kind of company you sell to. This page shows the profile your workspace runs on. The queue uses it to decide which companies are in your view and how well each one fits.

The profile is one sentence plus a set of filters. The queue filters on four of them right now: the industries, the size band, the locations, and the exclusions. Three more are recorded and shown but not scored yet: ownership, triggers and buyer roles. The page says which is which, so you never have to guess.

A workspace admin sets the profile on the setup screen (section 8.3) and can change it any time with the Edit button. Contributors and viewers read this page; they cannot change it. An admin can also keep the profile private, so that only admins see it (section 8.4).

8.2What each thing on the ICP screen means

• Who we are looking for. The profile in one sentence. Under it: the version (v1, v2 and so on, it goes up by one each time an admin approves), the date it was approved, and who approved it. If no profile has been approved yet the line reads "No ICP yet. Your workspace admin sets it on the ICP page."
• Attributes. The filters as plain lines: Industries, Size (for example 100 to 2,499 employees), Locations, Ownership, Triggers, Buyer roles. Ownership carries the note "recorded, not yet enforced": no company record in our data says who owns it, so it cannot filter anything yet. Triggers and buyer roles are shown for the same reason.
• Exclusions. What is left out: industries to avoid, a hard floor or ceiling on size, regions to avoid, and the free text you typed under the last question.
• Why this many. A short box built from your live numbers: "Your profile matches N companies we know about. M have evidence. K qualify. T are Pursue now." followed by one sentence that explains the Pursue now count, for example "2 accounts are Pursue now because they have evidence in three different areas, a strong signal and recent evidence." These numbers come from the same build that fills the Today queue, so they are never typed in by hand. The Today screen shows the same line under its tiles.
• Targeting rules. This card only appears when your workspace has rules of its own. Each rule shows an effect (a triangle for boost, a down arrow for downrank, a cross for exclude), a name and a plain description. They are shown, not applied to the score yet. Most workspaces have no rules of their own, so most people never see this card at all.
• Edit. Admins only. Opens the setup screen with everything filled in from the current profile. Approving there makes a new version.
• Your signals. A button next to Edit. It opens the page where your workspace chooses which kinds of event the desk counts (section 9).
• Open the library. A button to the Library, which is graded against this profile.

8.3Setting up who you sell to

The setup screen is at https://signals.srv1195619.hstgr.cloud/app/icp/setup. An admin sees it the first time they open a new workspace, before the queue, when the workspace has the setup gate on. Until an admin approves a profile, the other members see a page called Almost ready with the line "Your admin is setting up who you sell to. The queue opens when that is approved." and links to Team and Guide. Team, Guide, Manual and Troubleshooting always open, so the admin can invite the colleague who knows the profile, and everyone can read.

The screen is one question, then either a paste or six more questions.

1. Do you have an existing ideal client profile? Two big choices, Yes and No.
2. Yes. A box to paste your profile into. Press Read it. The app scans the words for industries, employee numbers, ownership words like PE or venture, countries, states and regions, buyer titles like CEO or COO, and the headings and sentences that say what a poor fit looks like, such as "poor fit", "red flags", "deal breakers", "out of scope", "avoid" or "exclude". It fills all six answers and the pickers for you to check: places you name become chips, places you rule out become places to skip, and one sentence about who buys is spread across all five buyer slots. Nothing is saved until you approve. No AI reads your text; it is a plain word scan, so check what it picked.
3. No. Six short questions, one at a time, with a Why it matters line under each. Previous and Next move you along. Your typing is saved as a draft as you go, so you can leave and come back. Under some questions sit pickers that feed the queue:

• Under Who gets the most value from it?: the industries (pick from the list, or Other with your own word), the size band (six presets, or type your own smallest and largest), locations (type one and add it), and ownership.
• Under What causes them to act now?: trigger chips such as new leadership, new funding, ownership change.
• Under Who buys, approves, uses, and can block the solution?: five slots: sponsor, budget, users, approvers, blockers.
• Under What makes a customer a poor fit?: disqualifier chips (too small, too big, industries to avoid, regions to avoid and so on) plus free text.

1. Review. The app writes your profile as one sentence in a fixed shape: "Our ideal customer is a (type of company) experiencing (problem), usually triggered by (event), with (sponsor) prepared to invest in (result). We should avoid customers with (disqualifiers)." The box is never empty: before you have answered anything it already holds the shape of the sentence, so you can see what you are editing. You can edit it, or leave it and the app keeps its own. What you approve is stored word for word, so the sentence on the ICP page is the one you read here. Under it, the filters as a plain list, for example "Companies in Technology with 100 to 2,499 employees, in the US".
2. Approve. The version goes up by one whenever something changed, the profile is stamped with your name and the date, and the queue is rebuilt right away with the new filters. Press Approve again with nothing changed and the version stays where it is, so a second press by mistake does not make a new version. No new web search runs; the app re filters and re scores what it already holds. The line under the button says it: you can change this any time from the ICP page.
3. Your signals. After Approve you land on the Your signals page (section 9). A band at the top says "Your profile is saved. Now choose what the desk watches; everything is on to start." with a Skip to your queue link. Every kind of event is on, so you can skip straight to Today and come back later. Switch off what you do not want counted, and the queue follows. Then a Next: Sources link in the band takes you to the last setup step, the Sources page (section 10).

If Approve shows a list of short lines instead, something in the profile cannot be used yet: an industry not on the list, a smallest size bigger than the largest, a pattern that does not work, or a sentence or answer that is too long. Fix each line it names and press Approve again. Nothing is saved until it passes.

What each role can do: admins set up, edit and approve. Contributors and viewers read the ICP page and never see the raw paste or the six answers, only the sentence and the filters. The Axsys team can load a profile for a workspace when you ask, for example the validated one a workspace was created with.

If the page gets stuck, see the ICP rows in Troubleshooting at https://signals.srv1195619.hstgr.cloud/help/troubleshooting, or ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai.

8.4When the ICP is private

On the Team page an admin has a switch, Members can see the ICP. It is on to start. When an admin turns it off, contributors and viewers see one line on the ICP page, "Your admin keeps the ICP private.", and nothing else of the document. The ICP item leaves their rail, and the starter set band on Sources goes with it, because that band names the first industry in the ICP (section 10.8). The queue still uses the profile exactly as before; only who can read it changes. Admins still see the whole page, with a small chip, "hidden from members", so they remember it is off. Section 13.5 shows the switch.

9.Your signals

9.1What the Your signals page is for

The desk watches for many kinds of event: a new CFO, a going concern warning, a wave of job posts. Your signals is the page where you choose which of those kinds count for your workspace. Every kind is on to start. Switch one off, and the desk stops counting that kind of event for you. Switch it back on any time.

Open it from the rail, from the Your signals button on the ICP page, or from Manage on Your signals on the Signal desk. The address is https://signals.srv1195619.hstgr.cloud/app/signals.

9.2What each thing on the Your signals screen means

Under the title, the line "What the desk watches for you, by category". If you have just come from the setup screen, a band at the top says "Your profile is saved. Now choose what the desk watches; everything is on to start." with a Skip to your queue link and a Next: Sources link. Everything is on, so skipping is fine. Next: Sources takes you to the last setup step (section 10).

The four tiles:

A search box narrows the rows to kinds whose name contains your word.

Then one section per category. The categories are the evidence areas from our catalog, seven today, with names like Capital & Finance, Jobs/HR & Employee Sentiment, and M&A & Integration. Each section starts with a header row: the category name, "N of M on", and, for admins and contributors, two buttons: All on and All off.

Under the header, one ruled row per kind of event:

Kinds with results in your view are listed first in each category. The rest sit behind a Show the other N link, so the page is not a wall.

The line at the foot says it in one breath: "Off means the desk stops counting that kind of event for your workspace. Nothing is deleted; switch it back on any time."

9.3What a switch does

Off means this: the next time your queue is built, every result of that kind is dropped before any scoring happens. So that kind never shows in the queue, in a brief, in the points, in the tiers, on the desk, or in the counts, including the why this many box. If an account had only that kind of evidence, it drops out of your queue. It is as if the desk never saw those events, for your workspace only.

On means the opposite, and on is how every kind starts.

Three things to know:

• It is per workspace. Your switches change nothing for any other workspace, and theirs change nothing for you.
• Off is not delete. The events stay in the platform. Switch the kind back on and they come straight back.
• It takes a moment. Your queue is rebuilt the next time a page loads after a flip, and a rebuild can take up to a minute. Refresh if you do not see the change yet.

9.4Who can flip a switch

Admins and contributors. Viewers see the switches greyed out and the line "Your admin or a contributor sets these". A read only workspace (the trial ended, section 3.2) cannot flip anything either. Every flip is saved with your name and the time, and a flip to off records the reason "switched off on the Your signals page".

9.5Your signals and the Library set

The Library (section 7) also has Keep in set and Exclude. Those shape your Library set, which is a different list from the kinds of event on this page, and the link between the two lists is not built yet. So today: Keep and Exclude change what the Library shows and counts, and the switches on Your signals change your queue. The Library card says this under its two buttons, so nobody has to guess.

9.6If the page gets stuck

• The switch will not move, and the line says "Your admin or a contributor sets these". You are a viewer, or the workspace is read only. Section 9.4.
• I switched a kind off and it still shows on Today. The queue is rebuilt on the next page load. Wait a moment and refresh.
• A category says "0 of M on". Someone pressed All off on that category. Press All on, or flip the ones you want.
• "This workspace is read-only." Your trial ended. Section 3.2.

10.Sources

10.1What the Sources page is for

A source is a place we read to find evidence. A newspaper is a source. A public filing is a source. So is a court record, or a list of building permits. Our collectors read these places on their own clocks, most of them overnight, and write down what they find about your accounts. That is where the results in your queue come from.

The Sources page lists the places we read for you, one row each. Every one is on to start. Switch one off, and its evidence stops counting for your workspace. Switch it back on any time.

Open it from the rail, or from the Next: Sources link on Your signals when you finish setup. The address is https://signals.srv1195619.hstgr.cloud/app/sources.

10.2What each thing on the Sources screen means

Across the top, a band of four tiles: how many channels are on, how many are off, how many results the channels that are on put in your queue, and the date of the newest result.

Then the block What Axsys collects for you. A channel is a group of feeds we read the same way. Google News and trade press is one channel. Inside it are 29 saved news searches, five trade sites we read directly and two press wires, 36 feeds in all. Each feed is one newspaper, one wire or one search. The nine channels today:

Three of the nine do not feed your queue the way the other six do, and each row says so in its own words. OSHA inspections is not collecting yet. SBA loan records and Financial profiles are loaded in batches when we load them, not on a clock, and Financial profiles is mostly there to fill in the numbers on an account. Everything we already found from all three is still in the platform, so they keep their row and their switch.

One ruled row per channel:

Under that block sits a second one, In the library, not collected yet. It lists the Library's other buckets: places we know about but do not read yet. A source your own workspace added (section 10.6) reads "Yours, queued, not collected yet", with its address, its industry, the date it was added, and a "behind a login" chip when you said so. There is no switch on those rows. They are there so you can see the whole picture.

Under that come the cards for adding sources of your own (sections 10.6 to 10.8) and the card What produced results (section 10.9).

10.3What a switch does

Off means this: the next time your queue is built, every result from that channel is dropped before any scoring happens. So that channel's evidence never shows in your queue, in a brief, in the points, in the tiers, on the desk, or in the counts. If an account had evidence only from that channel, it drops out of your queue. It is as if we never read that channel, for your workspace only.

On means the opposite, and on is how every channel starts.

Three things to know, the same three as on Your signals:

• It is per workspace. Your switches change nothing for any other workspace, and theirs change nothing for you. Axsys keeps reading every channel for everyone.
• Off is not delete. The results stay in the platform. Switch the channel back on and they come straight back.
• It takes a moment. Your queue is rebuilt the next time a page loads after a flip. Refresh if you do not see the change yet.

10.4Who can flip a switch

Admins only. Contributors and viewers see the switches greyed out with the note "Only an admin can change this." A read only workspace (the trial ended, section 3.2) cannot flip anything either; the note says the switches are locked. Every flip is recorded with who did it and the time.

This is stricter than Your signals, where contributors can flip too. A whole source is a bigger lever than one kind of event, so it belongs to the admin.

10.5One feed inside a channel

You can switch off a whole channel today. You cannot switch off one feed inside it. So you can turn off Google News and trade press, but you cannot turn off one newspaper in it and keep the rest.

Here is why, said plainly. When a collector writes down a result, it writes down the channel it came from, not the feed. So we cannot yet tell one newspaper's results from another's. Until the collectors record the feed, a switch per feed would be a promise we cannot keep. The Show the N feeds list is there so you can see what is inside a channel. It is not a set of switches. This is on purpose, not a mistake.

10.6Add your own source

You know places we do not read yet: a trade paper for your niche, a state permit register, a rival's news page. The card Add your own source lets you tell us. Admins and contributors can add; viewers can read the list.

Fill in:

• Name. What to call it, 2 to 120 characters.
• URL or feed address. The web address. It must start with http or https and be a real host name with a dot, like www.example.com/news. A user name or password inside the address is refused.
• Industry. Pick one from the list, or leave it blank.
• Notes. Up to 500 characters. What it covers, why it matters to you.
• It sits behind a login. Tick it when the site wants an account. A small box then asks What kind of login (up to 200 characters): a free account, a paid subscription, a customer portal. Say the kind. Do not type a password anywhere. The card says the same in bold. Axsys will handle the login with you.

Press Save. You see "Saved. Queued, not collected yet." and the row appears under In the library, not collected yet as "Yours, queued, not collected yet".

What queued means. Your source is written down, with your name and the date. Nothing reads it yet. No collector goes to that address tonight. Queued means it is in the line for the Axsys team to wire up, one source at a time, and until they do, it produces no results. This is the honest state, not a promise of a result next week. When a collector starts reading it, the row moves up into What Axsys collects for you with a switch.

Two rules. The same address cannot be added twice to one workspace; the second try says "duplicate". And the address may not point at a private network or at a local machine; that is refused with a plain reason.

One cap. Your whole workspace can write sixty times an hour, counting adding one, importing a file and applying a starter set together. A whole file counts as one. Past that the card says "Slow down a little." Wait a few minutes and carry on; nothing is lost but the press.

10.7Import a CSV

When you have many sources, the card Import a CSV takes them all at once. Up to 500 rows.

Make a spreadsheet and save it as CSV. The first line is the header. Columns, in any order: name, url, industry, notes, behind_login. One more is allowed: login_note. Name and url are required; the rest may be blank. Any other column is ignored. Put yes or true in behind_login when the site wants an account, and say what kind in login_note. Never put a password in the file. Every row gets the same checks as the form.

Pick the file and press Import. You see a count, "N added, M rejected", and a list of the rejected rows with the row number and the reason in plain words, for example "row 7: url: The address must start with http or https", "row 12: duplicate of row 4", "row 15: already in your workspace", or "row 501: over the cap". Rows past 500 are always rejected. The added rows arrive as queued, not collected yet, the same as a source added by hand (section 10.6). A blank line keeps its number and is skipped.

If nothing was added, look at the first rejected line. "file: missing column" means the header is wrong. Fix the file and import again; rows that already went in are skipped as "already in your workspace", so a second import does no harm.

10.8Start with the industry set

For each industry we know, the Axsys team is picking a set of sources worth reading, fifteen to twenty per industry. None of the sets has been loaded yet, so no workspace sees this band today. When a set does exist for your industry and you have not applied it yet, a band shows on the Sources page: "Start with the Food processing set: 18 sources picked for Food processing. They arrive as queued, not collected yet." with a button.

Press it. Every source in the set is copied into your workspace as your own rows, queued, with the note "From the Food processing starter set" and the one line reason each was picked. Sources you already have are skipped, so pressing it twice adds nothing. The band then goes away.

Your industry is the first one in your ICP, or the industry Axsys set on your workspace. When no set exists for it yet, there is no band. The sets are being reviewed one industry at a time, so a band may appear later. Admins and contributors can apply a set. One more case: if your admin keeps the ICP private (section 8.4), the band is for admins only, because the band names the industry from the ICP.

10.9What produced results

The card What produced results answers one question: which channels are earning their place. Two chips at the top, 7 days and 60 days. Pick one.

One row per channel, best first:

What the numbers mean, plainly. Results is how much a channel found. Accounts is how widely. A channel with 40 results on 2 accounts is loud about two companies; a channel with 12 results on 12 accounts is spread wide. The 7 day view says what is working now. The 60 day view says what works over a season, and it is the better one for deciding whether to switch a channel off.

One thing this card does not say: whether a channel ran at all. Zero results can mean "it ran and found nothing for you" or "it never ran". The line at the foot of the card says where that lives: "Whether a channel ran at all, and why not, is on the Health page per collector; per feed after feed level tracking lands." Chapter 11 is the Health page.

10.10Sources and Your signals, and where to go next

They answer two different questions. Your signals (section 9) asks which kinds of event count: a new CFO, a lawsuit, a wave of job posts. Sources asks which places we read for evidence. A result counts only when both its kind and its channel are on.

If you came here from setup, you are done: ICP, then Your signals, then Sources. Go to Today (section 4) and read your queue. You can come back to this page any time from the rail.

10.11If the page gets stuck

• The switch will not move, and the note says "Only an admin can change this." You are a contributor or a viewer. Section 10.4.
• I switched a channel off and its results still show on Today. The queue is rebuilt on the next page load. Wait a moment and refresh.
• I want to switch off one newspaper, not the whole channel. Not possible yet. Section 10.5.
• A row shows 0 results and an old last result date. That channel has found nothing new for your accounts. Not an error. Health (chapter 11) says whether it ran.
• "duplicate" when I add a source. That address is already in your workspace, maybe under another name. Look under In the library, not collected yet.
• My import says "N rejected". Read the reasons under the count; each names the row and the field. Section 10.7.
• I added a source a week ago and it has no results. Queued means not collected yet. Section 10.6 says what that means.
• "This workspace is read only, so the switch will not move." Your trial ended. Section 3.2.

11.Health

11.1What the Health page is for

The Sources page tells you what a channel found. Health tells you whether the collector behind it ran at all, and when it did not, why. It is the page to open when a channel looks quiet and you want to know if quiet means "nothing happened" or "it broke".

Any member can open it, from the rail. The address is https://signals.srv1195619.hstgr.cloud/app/health.

11.2What each thing on the Health screen means

At the top, three chips: 1 day, 7 days and 30 days. Pick the window you want to look at. Seven days is on when you arrive.

Our automation server only keeps about a fortnight of run history. When the chip you press reaches further back than that, the page uses the shorter real window and says so in a line under the chips, for example "The run history only goes back 14 days, so this window shows the last 14 days, not 30 days." Every count and every sentence on the page is then written against that shorter window, so no number claims a month it cannot see.

Then four tiles:

Then one ruled row per collector. Each row has a title, one plain sentence, the last run stamp (the date and time it last ran, or "No run in the window"), and a small code chip.

The title is the collector's own name in plain words: News sweep, Hiring and workforce sweep, Austin permits pull, and so on. Our internal name for the job never leaves our server. Every collector has its own name, so no two rows read the same. The Channel line under the sentence says which channels that collector feeds, in the same words the Sources page uses, and that is where a channel name appears.

The sentence is the part to read. Examples:

• "It ran and found results for one of your companies." Good. That is a collector that fired.
• "It ran, but found nothing for your companies." Also fine on a quiet day. The sources had things, none about you. Not broken.
• "It did not run in this window." This is the one to notice. Nothing was read.
• "It could not sign in to the service. The key or password needs a fix." Broken on our side; Axsys fixes it.
• "It started but never finished." A run that stalled.
• "It ran 4 times in the last 7 days and 2 failed. The last run: it ran and found results for one of your companies." A row counts its failures out loud, even when the newest run worked, so the Failed tile always has rows behind it.
• "We load this one by hand rather than on a schedule, so there is nothing to watch here yet." Nothing is wrong. No job runs behind that row, so there is no run to report.
• "We mostly use this one to fill in the numbers on your accounts. Very few of its finds reach your queue as results." Also nothing wrong. That one is doing background work rather than filling your queue.
• "This one is not collecting yet, so nothing from it has reached you." Honest, and ours. Nothing that collector was meant to read has ever reached you.
• "The automation behind this one is not switched on yet." Ours. The job named for that row is not on our automation server right now.
• "We ran out of time before we got to this one. Reload in a moment." The page gives itself 20 seconds to read our automation server, and it did not reach this row in time. Every other row is still true. Reload and it fills in.

The code chip is the same sentence in one word, so you can quote it to us. The chip prints it in capitals with spaces where the code has underscores, so ran_found shows as RAN FOUND, did_not_run as DID NOT RUN, failed_auth as FAILED AUTH, not_collecting as NOT COLLECTING and not_client_facing as NOT CLIENT FACING. To look one up in the manual, Appendix B, put it back the way the code is written: make it small letters and turn each space into an underscore. Holding your pointer over the chip shows the exact code too. Green means it found something for you, blue means it ran, red means it failed, amber means it is running right now, and plain grey means a quiet state: it did not run, or there is no job behind the row, or we ran out of time before reading it.

11.3What "fired" means

A collector fired when three things are all true: it ran, it found something, and what it found was about one of your companies. All three. "It ran" alone is not "it fired". A collector that ran and found nothing for you is not broken, it is quiet. A collector that did not run is the one to ask about.

11.4One row per collector today

One thing about the rows is true today and will change.

Rows are per collector, not per feed. The line at the foot of the page says it: "Rows are per collector until feed level tracking lands; then per feed." A collector like Google News reads 36 feeds. Today it writes down that it ran, not which of the 36 answered. So Health can say the whole collector ran and found nothing for you; it cannot yet say that one newspaper feed died. When the collectors record each feed, the page changes to one row per feed, grouped under the channel, and the Sources page can switch one feed off. Same reason as section 10.5.

11.5If Health gets stuck

• The page says "Health needs the n8n key on the server." Health reads our automation server, and this copy of the app was not given the key for it. Nothing is wrong with your workspace. Tell the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai; the Axsys team adds the key and the page fills in.
• "Health could not reach n8n right now." Our automation server did not answer. Wait a few minutes and reload.
• A row says "We ran out of time before we got to this one. Reload in a moment." The page gives itself 20 seconds to read our automation server and did not get that far. Nothing is broken. Wait a moment and reload; the answer is only held for half a minute, so the next one is a fresh read.
• A row says "We load this one by hand rather than on a schedule." There is no job behind that row to watch. Sources is where you see what it found.
• A row says "We mostly use this one to fill in the numbers on your accounts." That one runs in the background and rarely puts a result in your queue. Nothing to fix.
• A row says "This one is not collecting yet." That collector has never collected anything. It is ours. The Sources page says the same thing on the channel it belongs to.
• A row says "The automation behind this one is not switched on yet." That one is ours too. Tell the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai if it stays that way.
• The 30 days chip shows a shorter window. A line under the chips says so, for example "The run history only goes back 14 days, so this window shows the last 14 days, not 30 days." Our automation server only keeps about a fortnight of history. Nothing is broken and no number is hidden; the page is telling you what it can really see.
• A row says "No run in the window" for 1 day, but ran fine on 7 days. Some collectors run once a week. Widen the window.
• Every row says it ran, but a channel on Sources shows no results. Then the collector is fine and there was nothing about your companies. That is section 11.3, not a fault.

12.Send leads to your CRM

12.1Three ways out

When you decide an account is worth acting on, you can take it with you. Three buttons sit on the brief (section 5.2) and in the Today drawer (section 4.3):

• Download CSV. One lead as a spreadsheet row: a header line and one row, 71 columns. Opens in Excel as is. Any member can download.
• Download JSON. The same lead as a JSON file, for a developer or a tool that reads JSON. Any member can download.
• Send to CRM. Sends the same lead straight to your CRM through a webhook address your admin saved on the Team page. Admins and contributors can send. A viewer sees the button too, greyed out, with "Only an admin or a contributor can send." on it when you hold your pointer over it.

The files are named axsys-<workspace>-<account>.csv or .json. Every download is written to the audit log with your name.

12.2What is in a lead

All three carry the same lead, in plain words:

• Who sent it. The version of the shape (always "1" today), the time, and your workspace's name.
• The company. Its id, name, website domain, headcount, industry and head office. When we do not know one, the field is empty, never missing, so a mapping never breaks.
• The scoring. The score, the tier and its status word, the segment, fit, timing, the confidence label, and one line of why. That line is the engine's own short labels with their points, like "Size band fit (100 to 2,000 employees) 25 of 25.", one after another. It is the same arithmetic as the Why this score card, not the plain sentence at the top of that card; the sentence stays on the screen for now.
• The top signal. The newest piece of evidence: its title, date and excerpt.
• The evidence. Up to eight results, newest first, each with its title, date, channel, a link when we have one, and the excerpt.
• A link back to the brief in Axsys Signals.

There are no people in a lead yet. Contacts come later. The exact field list, with an example, is in the file docs/signals/crm-payload-v1.md in the product repository; ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai for a copy if your CRM person needs it.

12.3Sending to your CRM

Press Send to CRM. The button waits up to eight seconds for your CRM to answer, then tells you one of two things:

• "Sent to your CRM." It arrived. Your CRM answered that it took the lead.
• "The push failed: (the reason). Nothing is queued; try again." It did not arrive. The reason is what your CRM said, or "their server said 404" and the like, or that no answer came at all.

That second line is the whole rule for failures: there is no retry. A failed push is not saved for later and is not tried again in the night. Fix the cause (the address changed, the Zap is off, the trigger is paused) and press the button again. The Team page shows the last push time and its status so an admin can see the same thing.

A workspace can send sixty leads an hour. Past that the button says "Slow down a little."

When the button is grey with the note "No webhook set. An admin sets it on the Team page.", nobody has saved an address yet. Section 12.4.

12.4Setting up the webhook on the Team page

Admins only, on the Workspace settings card of the Team page (section 13.5).

1. Your CRM person gives you a webhook address, the place your CRM listens for leads. Paste it into the Webhook url field. It must start with https and be a real host name.
2. Press Save. A box appears with a long secret and the words "Copy it now. It is shown once." Press Copy and hand the secret to your CRM person. Close the page and it is gone from the screen for good; the page only ever shows its last four characters after that. If it is lost, save the address again and a new secret is made.
3. Press Test. A small test message goes to the address. You see "Your CRM answered 200. The test worked." or "The test failed with (status): (reason)."
4. Open any brief and press Send to CRM. The first real lead arrives.

Remove clears the address and the secret; the buttons on the brief go grey again. Under the field is a line, "Last push (date): (status)", or "No push yet."

Addresses that are refused, with the reason under the field in plain words: not https, no dot in the host name, an IP address instead of a name, a user name or password inside the address, a host name that does not resolve, or an address that points at a private network, a local machine, or back at Axsys. We check the address when it is saved and again at every send.

12.5The signature, in plain words

Every message we send carries a signature so your CRM can be sure it came from us and was not changed on the way. Here is the idea. We take the time we sent it, a dot, and the exact message, and run it through a one way math function together with your secret. The answer is a string of 64 characters. We put that string in a header called X-Axsys-Signature, and the time in X-Axsys-Timestamp. Your side does the same math with the same secret. If the two strings match, the message is real and untouched. If the time is more than five minutes old, throw it away, so nobody can replay an old message.

Most no code tools, Zapier and HubSpot workflows among them, cannot do this math. For them, the address itself is the secret: the long random part of the address is what keeps strangers out, so do not post it anywhere public. A team that runs its own small service in front of the CRM can check the signature there.

12.6The recipes

We wrote three step by step recipes: Zapier into HubSpot, HubSpot's own webhook trigger, and Salesforce by Zapier or by a Flow. They are in the file docs/signals/crm-recipes-2026-09-05.md in the product repository; ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai for a copy. One honest line about them, printed at the top of the file too: they were written from the vendors' documentation and have not yet been tried on a trial account. When you try one, tell us what did not match and we fix the page.

12.7If a send gets stuck

• The button is grey: "No webhook set. An admin sets it on the Team page." Section 12.4.
• "Only an admin or a contributor can send." You are a viewer. Download the file instead, or ask a teammate.
• "The push failed: their server said 404." The address is wrong or was regenerated. The admin saves the new one and you press again.
• "The push failed: no answer." Your CRM did not reply in eight seconds. Check the address and the tool behind it, then press again. Nothing was queued.
• "Slow down a little." Sixty sends in the hour. Wait.
• "Too many sends are in flight right now." The whole app was busy sending at that second, which is rare. Nothing was queued and nothing was lost. Wait a moment and press again.
• The Test worked but a real lead fails. The test message is small; the lead is bigger. Ask your CRM person whether the tool caps the message size or refuses a field.

13.Team

13.1What the Team page is for

The Team page shows who is in your workspace and what each person can do. Admins invite people here, change roles, remove people, and set the two workspace settings: whether members see the ICP, and the CRM webhook.

13.2What each thing on the Team screen means

• Invite a teammate. Admins only. A work email, a role, and a Send invite button. The role choices are Viewer (sees the queue and library), Contributor (rates signals, shapes the set) and Admin (manages people, workspace settings and the ICP). Section 16 has the full table. Admins set up and edit the ICP on the setup screen (section 8.3).
• Members. One row per person: initials, name, a You chip on your own row, a Gold owner chip on one person, the email and the join date, the role, and for admins a Remove button.
• Pending invites. Admins only. Invitations sent but not yet accepted: the email, the role, when it was sent, when it expires, whether it went out by email or was shared by hand, and a Revoke button.
• Workspace settings. Admins only. Section 13.5.

The gold owner is the first admin who accepted an invitation to this workspace; it is the one account that can never be removed without a handover to someone else, and that is all you need to know about it.

13.3Inviting someone

1. Type the person's work email.
2. Pick a role. Section 16 says what each can do.
3. Click Send invite.

You see "Invite emailed to (address)". The link in that email works for 14 days. If our mail could not go out, you see "Invite created. Email did not go out from here, so share this link" with the link to copy and send yourself.

Only one live link exists per email per workspace. Sending a new invite cancels the old link. If you see "That person is already a member", they are in; nothing to do.

13.4Changing a role or removing someone

Admins see a dropdown in the role column of every row but their own. Pick a new role; it saves at once and shows "Role updated". You cannot change your own role. If you try to demote the last admin you see "That is the last admin; promote someone else first."

To remove someone, click Remove on their row and answer "Remove (name) from this workspace?". You see "Removed" and they lose access at once. Your own row and the gold owner's row have no Remove button: you cannot remove yourself, and the gold owner needs a handover first.

Every invite, role change and removal is written to the audit log with who did it and when (section 15).

13.5Workspace settings

A card admins see under the members list. Two things live on it.

Members can see the ICP. One switch, on to start. Under it: "Off keeps the ICP for admins only. The queue still uses it." Turn it off and contributors and viewers see "Your admin keeps the ICP private." on the ICP page, and the ICP item leaves their rail. You see "The ICP is now for admins only. The queue still uses it." and a small "hidden from members" chip next to the switch. Turn it back on and you see "Members can see the ICP again." Nothing about the queue or the scoring changes either way. Section 8.4.

The CRM webhook. A Webhook url field, then Save, Test and Remove, and a line with the last push. Save shows the secret once, with "Copy it now. It is shown once." Chapter 12 walks through it step by step (section 12.4). Until an address is saved the line reads "No webhook yet. Paste the url your CRM gave you and press Save."

In a read only workspace (the trial ended) the card shows but everything on it is disabled.

13.6If the Team page gets stuck

• I do not see Invite a teammate, or the Workspace settings card. You are not an admin. Ask one.
• The invitation email never arrived. Check spam. Then ask your admin to look at Pending invites; the link is there to share by hand, or they can revoke and send again.
• The webhook Save is refused. The reason is printed under the field. Section 12.4 lists them.
• "This workspace is read-only." Your trial ended. Section 3.2.

14.Optional deep discovery

Two more pages exist for teams that want the long questionnaire: Discovery and Workshop. They are reachable at https://signals.srv1195619.hstgr.cloud/app/discovery and https://signals.srv1195619.hstgr.cloud/app/workshop.

Discovery is a long questionnaire in two parts. Part 1 covers your ICP and problems. Part 2 walks through the kinds of events, one category at a time. Answers save as you type. Admins and contributors can start it and answer. You can skip any question.

Workshop is where your answers and the desk's draft answers sit side by side. Its tiles count questions with any answer, questions both sides answered, resolved, and still to decide. An admin picks yours, the desk's, or writes a merged final for each question.

Neither is required. Nothing in the queue depends on them. You can run Axsys Signals for months and never open them. That is on purpose: the profile the queue uses comes from the ICP page and the setup flow, not from here. Discovery and Workshop are kept out of the left rail for the same reason; the addresses keep working, and both pages say "Optional deep discovery" under their title.

15.What happens to what you rate and type

Here is where each thing goes, plainly.

• Good lead and Bad lead, with the reason and the why are saved per workspace, per lead, with your name and the time. Your teammates see the tally on the desk and the brief, like "Team: 2 good, 1 bad"; no names are shown there. People in other workspaces never see them. The Axsys team reviews them weekly and decides, by hand, whether a rule should change. Nothing in the scoring changes by itself. Not overnight, not after a hundred votes. If you ever see an older line saying ratings feed back into scoring or train the model, it is out of date; what is written here is the truth.
• The 1 to 5 stamps and your why notes on the desk's source library are saved the same way, per kind of signal, and shown to your team as an average and a count.
• Keep in set and Exclude are saved per workspace with your name. Today they change the Library, not the queue (section 7.4).
• The switches on Your signals are saved per workspace with your name and the time. A flip to off records the reason. They change your queue and nobody else's (section 9).
• The switches on Sources are saved per workspace with your name and the time. Only an admin can flip one. They change your queue and nobody else's (section 10).
• Sources you add, import or take from a starter set are saved as your workspace's own rows, with your name and the date. Nobody outside your workspace sees them. Nothing collects from them until Axsys wires each one up (section 10.6).
• Downloads and sends to your CRM are each written to the audit log with your name, the account and, for a send, the answer your CRM gave. The lead itself goes only where you sent it. The webhook secret is stored for your workspace and never shown again after the one time (section 12.4).
• Reports from the help button go to the Axsys team's issue list with your email, your workspace and the page you were on, and are reviewed on Thursdays (section 3.6).
• The tour records only that you finished or skipped it, so it does not start again.
• Discovery answers are saved per workspace and shown to your team on the Workshop page. Nothing reads them into the queue.
• The audit log. Every password reset, invite, role change, removal, vote, stamp, note, set change, signal switch, source switch, source added, setting changed, download, send, report and access request is written to a log with who, when, what, and the network address it came from. It cannot be edited. There is no screen for it yet; an admin can ask the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai for an extract.

Who can see what, in one line each:

• Everyone in your workspace sees the same queue, the same brief, the same library, and each other's ratings as a team average and a count, not by name.
• Nobody outside your workspace sees any of it.
• The Axsys team can see it, to run the weekly review and to help you when you ask.

16.Roles

Every person in a workspace has one of three roles. Whoever invited you picked yours. Your role controls which buttons you see. It does not make your ratings count less.

If you expect a button and it is not there, it is almost always your role, not a mistake.

17.Frequently asked questions

Why is there only one account in tier one?

Tier 1 needs a score of 80 or more, and fit alone tops out at 15, so an account needs strong fit (evidence from several different areas, the right size, a complete profile) plus timing points. Very few companies have all of that at once, so the list is honest rather than long; it grows as more sources are read and more companies are added. You no longer have to take that on trust: open any account and the Why this score card names the parts it is short on and how many points each one is worth (section 4.3).

One account is not Tier 1 and I want to know why.

Open it. The sentence at the top of the Why this score card says where it sits and what it loses points on, every row shows what that part earned against what was on offer, and the last line says how many points would move it up and which part has the most room. On Today you can see the worst part without opening anything: it is the "Biggest gap" line under the account name.

Can I make a score go up?

Partly. More evidence arrives on its own as we collect. A timing estimate waits on a real dated event. What we know about the company fills in as we collect, and you can tell us what we are missing. Your size band is the one thing an admin can change, and only if it is wrong. Section 5.3 goes through all four.

Where is the detail on an account?

Click the row on Today; the drawer shows the score breakdown and the timeline. Click Open the full brief for the full page with the signal mix.

What is Discovery and do I have to do it?

Discovery is an optional long questionnaire for teams that want to go deep on their profile and signals. You do not have to do it; nothing in the queue depends on it.

I tried to sign in many times and it stopped working.

Ten wrong tries lock your email for up to an hour. Reset your password from the sign in page; that clears the lock and signs you in.

Why does a result say Strongly Supported?

That kind of signal usually shows up in more than one independent stream at once, so our catalog labels it that way. Today the label describes the kind of signal, not a check of that one result, so read it as a hint and look at the evidence line.

Where do my ratings go?

They are saved in your workspace with your name, and your team sees them on the desk. The Axsys team reviews them weekly; nothing in the scoring changes by itself.

Can I stop the desk counting a kind of event?

Yes. Open Your signals and click the switch on that row. Off means every result of that kind is dropped before scoring, for your workspace only. Nothing is deleted; switch it back on any time. Admins and contributors can flip switches. Section 9.

Can I stop counting a whole source, like Google News?

Yes, if you are an admin. Open Sources and click the switch on that row. Off means every result from that channel is dropped before scoring, for your workspace only. Nothing is deleted, and Axsys keeps reading it for everyone else. One newspaper inside a channel cannot be switched off yet. Section 10.

Can I export a lead to my CRM?

Yes. On the brief or in the Today drawer: Download CSV, Download JSON, or Send to CRM. Send to CRM needs a webhook address, which an admin saves on the Team page. There is no retry: a failed send says so and you press again. Chapter 12.

A channel shows no results. Is it broken?

Open Health. If its row says it ran and found nothing for your companies, it is quiet, not broken. If it says it did not run, or failed, that is ours to fix. Chapter 11.

Can I add a source you do not read?

Yes, on the Sources page: one at a time, or 500 in a CSV, or the starter set for your industry. They arrive as queued, which means written down and not collected yet. Never type a password. Section 10.6.

Where did the tour go, and how do I see it again?

It starts by itself once. To run it again, press Take the tour at the top right of any signed in page, or pick the same words from the Help menu beside it. Section 3.5.

I have no invitation. How do I get in?

Go to https://signals.srv1195619.hstgr.cloud/request and ask. we write back within two working days. Section 2.7.

Can I ask the product a question?

Yes. Press Help at the top right of any signed in page and choose Ask Signals. A panel opens on the right; type your question in plain words. It answers about your own workspace only, from its own data: your queue, an account and its score, results, sources and switches, the ICP unless your admin keeps it private, whether a collector ran, the Library and this guide. It changes nothing and it never guesses a number. Twenty questions an hour per person, sixty a day per workspace. Section 3.7.

The page says One moment, or Live data is delayed.

Our data provider is slow. Nothing is lost. Wait a minute and try again, and check https://signals.srv1195619.hstgr.cloud/status if it keeps happening.

18.What is coming next

Most of the earlier list is done. The nightly run came first, one release ago: every morning at 06:30 UTC the app carries the night's findings in and scores every company again, so the queue moves by itself and nobody has to start it. This release adds the rest: the ICP setup flow, Your signals, Sources with your own sources and the starter sets, What produced results, Send to CRM, per lead Good and Bad with a reason, the Source link on results, honest last result dates on the desk, Health, the first login tour, the help button with Report a problem, the ICP privacy switch and Ask for access. The chapters above show each one.

In the order we expect them, with no dates promised:

1. Per feed on Sources and Health. A switch for one feed inside a channel, like one newspaper, and a Health row per feed, once the collectors write down which feed each result came from (sections 10.5 and 11.4).
2. Collecting from your own sources. Today a source you add is queued and nothing reads it. Wiring them up, and handling the ones behind a login with you, comes next.
3. Retry on a failed send, and native HubSpot and Salesforce connectors. Today a failed push is one you press again.
4. People on the lead. Contacts on the brief and in the CRM payload, once the privacy decision is made.
5. The industry starter sets. The sets of sources picked for each industry are still being reviewed, so no workspace can apply one yet (section 10.8).

Also on the list, said out loud because this guide mentions them: scoring on the ownership, triggers and buyer roles you record in the ICP, the Library's Keep and Exclude reaching the queue once its list is linked to the desk's catalog (the switches on Your signals and Sources already do, sections 9 and 10), and a screen for the audit log.

19.Getting help

• The fastest way. Press Help at the top right of any signed in page and choose Report a problem. Say what you did, what you expected and what you saw. Reports are reviewed on Thursdays (section 3.6).
• Something is wrong or confusing. Write to the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai. Say which screen you were on and what you saw. The exact words on the page help most.
• Is it me or is it down? Open https://signals.srv1195619.hstgr.cloud/status. It needs no sign in. It says whether sign in and data are Normal or Delayed, and carries any note from the team.
• A message I do not understand. The troubleshooting page lists every message the app can show, what it means and what to do: https://signals.srv1195619.hstgr.cloud/help/troubleshooting.
• Account and access questions. Your workspace admin can see who is in the workspace and resend an invitation. Anything beyond that, the person who invited you, or the Axsys team at jonathan.bayo@axsys.ai.

End of User Guide.