ShipsAt launchWave 2Wave 3Wave 4
Build your own

Describe the work. It’s done by morning.

Every revenue team runs on work nobody has the hours for — chasing the signal, drafting the follow-up, checking the account before the call. Build an agent to do that part. It works through the night, reacts in hours instead of quarters, and hands you the decision.

The work that waits

The signals arrive. Nobody has the hours to chase them.

The churn signal
Usage at a big account halves in March. It sits in the system, visible to anyone who goes looking.
The catch
Nobody went looking until July.
The expansion signal
Three teams at one account hit their seat limit inside the same fortnight.
The catch
You heard about it at renewal.
The people
They know exactly what to do about both. They are genuinely good at this.
The catch
They carry forty accounts each.
Growth you already had.None of this was a knowledge problem. Every signal was sitting in the system and every person knew the play. There were never enough hours in anyone’s week to run it.
Building one

Describe the job. Beacon builds the worker.

Beacon agents come with every module you switch on. This page is about the other kind — your own agents, the ones you build because only your company needed them.

No code, no analyst, no ticket into anyone’s backlog. You tell Ask Beacon what you want handled, and it comes back with an agent that handles it — fences already on.

Beacon.Ask Beacon
I want to catch enterprise accounts going quiet before the renewal, not after.
I can run that as an agent — watching usage across your EMEA enterprise accounts, flagging a drop big enough to matter for the renewal. Should it draft the outreach too, or only flag it?
Draft it too, and send it to whoever owns the account.
Done. Here’s your agent — it starts tonight.
Quiet-account watchHeld for your yes
WatchesUsage falling more than 20% over sixty days.
Works onEMEA enterprise accounts — the ones you already own.
DraftsOutreach to the account owner, with the renewal at risk and the reason attached.
ThenWaits for a person to approve it. Every time.
Runs nightly · live in 4 minutes · nothing written in code
Ask a follow-up…
Four minutesFrom the sentence you typed to an agent that is actually running.
Every nightIt doesn’t forget, doesn’t get busy, and doesn’t take a holiday.
Nothing to maintainNo spreadsheet beside the system. No script only one person understands.

Illustrative. Beacon builds the agent by asking — there is no blank canvas, no formula, and no engineer.

The next morning

It went to work while the office was shut.

That one sentence bought a night shift. Here is what the agent you just built actually did with it, before anybody sat down.

03:12

Checked all 214 enterprise accounts. Three had gone quiet.

03:14

Pulled the renewal date, the value and the owner for each one.

03:15

Wrote three drafts. Three reasons, three different asks.

07:40

All three waiting for the account owner, reasoning attached.

09:05

She approved two over coffee and rewrote the third.

Four months of warning.€310k of renewal caught while there was still time to do something about it — for twelve minutes of somebody’s morning. The old version of this was a quarterly review, and a surprise.

Illustrative. Every draft waited for a person. Nothing was sent on its own.

Smarter than the analyst you can’t hire

Your best analyst sees one team. Your agent sees the whole journey.

An agent you build here isn’t stuck in the department that built it. It reads the same lifecycle everything else runs on — so when it looks at a renewal it already knows which campaign won the account, how onboarding went, and what the customer actually adopted. That is the part a person can’t do, however good they are: nobody holds all five stages in their head at once.

Three agents you built, one lifecycle

 

Same journeys, same definitions, same numbers your people read. Nothing your agent works from is a second version of the truth — which is why it can be trusted to act early instead of only reporting late.

See the lifecycle it reads →

Where the capacity goes

Everything nobody has time for, done every night.

Each of these was described in a sentence by the person who needed it. None of them went near an engineer.

Who builds itWhat it doesWhat they get back
Customer success leadFlags accounts going quiet before the renewal window opens, with the reason.Four months of warning instead of four weeks.
Head of salesChecks every open deal against how deals like it have actually closed.Forecast calls made on evidence, not optimism.
Revenue operationsTidies the records underneath the forecast every night and proposes the fixes.A Monday pipeline nobody has to apologise for.
Marketing leadWatches which segments convert and where spend has stopped paying back.Budget moved this month, not next quarter.
FinanceDrafts the variance note the moment the month closes, with the drivers named.Close week, minus a day.
Support leadSpots the ticket pattern that ran ahead of your last three cancellations.Saves found while they were still saveable.

And it counts. Your agents are capacity in the plan.

Build one and it doesn’t sit outside the system as somebody’s clever side project. It lands in the same capacity number the team is planned on — beside your people, beside Beacon agents — and it gets checked against what the plan actually needs before anyone commits to that plan.

People Beacon agents Your own agents What the plan needs
Customer success 106% covered
Sales operations 101%
Finance operations 92%

Customer success is covered because somebody built the difference. Finance operations isn’t — nobody has built one there yet. That gap is a hire, or a sentence.

Illustrative. Beacon shows you the split before you add a role. It never switches an agent on, and it never makes the hire.

See the capacity system →

The part you don’t have to worry about

It suggests everything. It decides nothing.

  • It can’t produce a number. The engine computes every figure. Your agent explains one, and can’t overrule it.
  • It can’t see more than you do. It works inside your access. When yours narrows, so does its.
  • It can’t delete anything. Not a record, not a field, not a note. There is no version of this it’s allowed to do.
  • It can’t touch your billing or your books. It reads them. That’s the whole permission.

You didn’t write a single one of those, and you can’t switch one off. They aren’t settings on your agent. They’re the shape of what an agent is here — which is why more capacity doesn’t have to mean more risk.

The same fences as ours. Held tighter.

Beacon agents can, over time, earn the right to apply a small reversible change without asking. Your own agents can’t — not until an admin grants it, to that one agent, after it has earned it. Agents you build get less rope than the ones we ship, not more.

And a short list of calls is closed to every agent for good — committing the plan, moving a target, authorising a capital event. Those stay with the people whose names are on them.

Where it fits

Beacon agents come with the modules. Your own agents do the rest.

Beacon agents arrive with every module you switch on and work when nobody asked. Your own agents take the job that only your company has. Ask Beacon answers when you ask it, and builds the second kind for you. Three jobs, one set of fences, one capacity number.

Beacon agents and the fences ship at launch. Building your own follows in wave two.