> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pavoai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Forward-Deployed Scientist

> A Pavo scientist embeds in your team, runs the improvement loop with you until a metric moves, then hands it back.

<Note>
  **Everything in self-serve, plus a scientist who runs it with you.** A Forward-Deployed Scientist is a builder who embeds in your team and works in your production environment, not an advisor, not a consultant writing decks. They run the improvement loop with you until a metric moves, then hand it back so your team runs it.
</Note>

**Designed for** teams that want a metric win *proven* before they invest their own people.

## How the work happens

**System Knowledge**: In weeks 1–2 your scientist builds knowledge-book v0 *with* you, ingesting the tribal knowledge that lives in your team's heads.

**Agent Studio**: Your scientist formulates hypotheses, generates multiple approaches, and tests them in parallel. You see the plan before anything ships.

## The 8-week loop (single metric)

| Stage                | Weeks | You                                  | Leaves behind                       |
| -------------------- | ----- | ------------------------------------ | ----------------------------------- |
| Define success       | 1     | Agree the win + how it's measured    | Metric definition                   |
| Assemble context     | 1–2   | Grant access, share tribal knowledge | Connected stack · KB v0             |
| Formulate hypotheses | 3–4   | Pressure-test them                   | Hypothesis set · offline harness    |
| Build & evaluate     | 3–4   | Review the first offline read        | Offline→online correlation          |
| Productionize        | 5–6   | Approve the live test                | Live A/B · checkpoint               |
| Measure & hand over  | 7–8   | Take handover                        | Metric moved · self-sufficient team |

Two metrics (e.g. pricing **and** recommendation) extend the same loop, typically 12 weeks.

## How the handback works

Your team doesn't watch from the sidelines. Across the engagement they move from **observers → co-builders → operators**: shadowing the first tests, then driving them with your scientist alongside, then running the loop on their own. You're left with the moved metric *and* the people and playbook to keep moving it.

## What we ask of you

Self-sufficiency is real work, not a switch we flip on the last day. It needs you to name the people who'll own the loop and give them time on it during the engagement.

## This is bounded, not a services contract

The engagement is time-boxed and self-liquidating. It sits on the Pavo platform - reusable primitives, not bespoke code only we can maintain - so when we leave, nothing breaks. A healthy engagement needs *less* of us over time, not more.

## What stays yours

The same isolation as self-serve - a per-client service account, your code and data never leaving your setup - plus a self-sufficient team at the end.

## How to start

One conversation to find the problem worth solving. [Book a demo](https://pavoai.com), no procurement gauntlet to begin.