Back

Pricing Customers

Five agents. Five desks. One shared memory.

The crew that runs your analytics is five AI agents, each owning one desk and doing one job well. They draw from the same memory of your business, so nothing is relearned — and one human editor signs every release before it reaches you.

SScout BBuilder AAnalyst KKeeper TTuner

Desk by desk

Five desks. Each with one job, done well.

Here is what each agent watches, what it produces, and what it hands to the next desk — with a real line from the field.

Agent 01 / 05
Context desk
Scout

Learns your business, so no one starts from a blank page.

Maps your business. Your KPIs, your definitions, your data sources — captured from a 30-minute brief and turned into a living context the whole crew works from.

Signals in
Your goals, key metrics and where your data lives — in plain language.
The work
Builds a map of what each number means: naming, rules, sources, the definitions that keep answers honest.
Hands off
A shared context every other agent reads from. Learned once, never re-asked.
Logged "Active client" excludes churned trials. Applied everywhere since — no team has had to re-explain it.
Agent 02 / 05
App desk
Builder

Shapes the app around your teams, not a template.

Assembles your app. Views, filters and indicators shaped around how each of your teams actually works — not a stock dashboard with your logo dropped on top.

Signals in
Scout's business map, plus how each team really uses its numbers day to day.
The work
Builds the views and layouts each team needs — bespoke, not boilerplate.
Hands off
A working app on your data, ready for the Analyst to fill with answers.
Shipped A sales view with margin-by-quote front and centre — the exact shape the sales team asked for, not the default.
Agent 03 / 05
Answers desk
Analyst

Answers through your context. No made-up numbers.

Runs the analysis. Questions in plain language, answered through your semantic layer and business context — so the numbers hold up and the risky ones reach a human.

Signals in
Plain-language questions from you and your teams, from inside the app.
The work
Answers via your semantic layer — never free-styling numbers — and flags high-stakes ones for review.
Hands off
Answers that hold up, with the ones that matter routed to your lead.
Answered "What was Q2 churn?" → 8.4%, traced to the exact filter and definition, ready to defend in a board meeting.
Agent 04 / 05
Operations desk
Keeper

Watches the pipes so you never have to.

Watches the pipes, 24/7. Data freshness, reliability and access rules. When something drifts, it's caught, flagged and fixed before it ever reaches your screen.

Signals in
Your live sources, pipelines and permission rules.
The work
Monitors freshness, reliability and access around the clock. Catches drift early.
Hands off
An app you can trust — and a heads-up to the Tuner when something needs a fix.
Caught A source went stale at 03:14. Flagged and backfilled before the morning standup. No one noticed a gap.
Agent 05 / 05
Corrections desk
Tuner

Turns your requests into daily releases.

Prepares the next release. Your requests and corrections, turned into the improvements that ship — around 20 a month — and fed back into the shared memory so nothing is lost.

Signals in
Change requests and corrections, in plain language, from inside the app.
The work
Turns them into shipped improvements and writes what it learns back to Scout's map.
Hands off
A daily release, and an app that compounds instead of resetting.
Shipped "Add churn risk by segment" asked on Monday. Live by Thursday, signed by your lead.

What makes them a crew

Five agents. One memory.

The difference between five AI tools and a crew is a shared context. Everything one agent learns, the others can use.

Point five separate AI tools at your data and you get five sources of truth, none of them talking. Ask the same question twice and you re-explain your business twice. The crew works the other way.

Everything Scout learns, everything the Analyst pins down, everything the Tuner corrects lives in one shared context every agent reads from and writes to. That is why the app compounds instead of resetting — and why you never explain the same thing twice.

The context carries from one release to the next, and from one agent to the next. Nothing is relearned.

The context ledger Shared · read + write
Scout "Active client" = paying, non-trial.learned wk 12 · applied to every view since
Analyst Q2 churn method pinned at 8.4%.definition locked · reused on every ask
Keeper Nightly source SLA set to 04:00.monitored 24/7 · alerts on drift
Tuner Margin-by-quote view shipped.signed by your lead · fed back to the map
Written once. Read by everyone. Never relearned.

The last word

The crew drafts. A human signs.

Five agents do the work continuously. None of them is the last link — that seat belongs to a person.

Your Toucan Lead
Human · signs every edition

Accountable for every number. Your lead reviews each release before it reaches you, signs what ships, and answers when something's wrong. Nothing goes to production without your go.

Not a support inbox, not a queue, not a ticket. A person you know by name — the editor who reads every edition before it goes out, and the reason five autonomous agents stay honest.

Five agents do the work. One human has the last word.

Put the crew on your data. First result in 48 hours.

Hire a Crew →

Rather build it yourself? Start free in Studio →