Forward-Deployed Engineer
Hand over a problem rather than steering the steps — a migration, ongoing ownership of one of your servers, or a bespoke automation built for your stack and handed back to you.
Overview
Forward-Deployed Engineer (route /fde) is for the times you'd rather hand over a problem than drive the swarm yourself. You describe the outcome you want; SnapFlow takes the whole engagement — planning it, running it, and proving it worked.
An engagement is still an ordinary mission underneath. It appears in Mission Control, pauses at the same approval gates, streams the same Flight Log, and settles under the same rule: you're only charged for work that was verified. What differs is that you're buying the outcome rather than steering the steps, and engagements are priced at a premium over the equivalent self-serve mission.
The three engagement types
Each card on the page opens a short form. Pick the one that matches the shape of your problem:
| Engagement | Use it for |
|---|---|
| Migration | Moving an app, server, or whole environment from one host or cloud provider to another. SnapFlow discovers the source, checks permissions on both ends before starting, provisions the destination, and verifies parity before any traffic cuts over. |
| Embedded Ops | Ongoing operational ownership of one of your own machines — an engineer that works inside your environment rather than provisioning cloud resources from outside it. Requires a connected server. |
| Custom Build | A bespoke automation built for your specific stack. Once the engagement is verified, the automation becomes a reusable recipe your team re-runs on its own. |
Migration
Name the source (a host or a provider) and the destination provider, plus anything that matters — a maintenance window, a database that must not lose rows, a service that has to keep answering. The plan comes back with a human approval gate before the destination is provisioned, and a second, separate gate before cutover. Cutover is always its own decision; approving the build does not approve the switch.
Embedded Ops
Choose one of your connected servers from the dropdown, then describe the work it owns. The dropdown lists only servers that are currently connected — if it's empty, connect one first under Infrastructure & Targets → the Connect a server flow. Commands run on that machine through the same agent and pass the same per-command safety gate as any other mission.
Embedded Ops is for work an engineer would keep owning on that box — keeping a service healthy, handling its upgrades, fixing it when it drifts. For a single task on a server, an ordinary mission from Mission Control is cheaper and does the same thing.
Custom Build
Describe the stack, the manual process you want automated, and what "done" looks like. SnapFlow discovers how your environment actually works and builds the automation against it.
The deliverable is the interesting part: when the engagement is verified, its proven steps are distilled into a parameterised recipe that shows up under Automations built for you at the bottom of the page. That recipe is yours — later missions matching the same goal reuse it, so the work you paid for once keeps paying off without another engagement.
Starting an engagement
- Open Forward-Deployed in the left-hand navigation.
- Click the button on the card that matches your problem — for Embedded Ops, pick the connected server first.
- Describe the engagement in the box. Be concrete about the outcome and any constraint that matters; this text is what the plan is built from.
- Click Start engagement. SnapFlow records the request and starts the mission.
- Follow it in Mission Control — the confirmation message links you there. Approvals appear in Triage & Gates as usual.
Tracking your engagements
Below the three cards, Your engagements lists everything you've requested, newest first, with its type, what you asked for, and when. Once a request has started, a View mission link takes you to that run's page.
Automations built for you lists the recipes produced by verified Custom Build engagements — each with how many steps it runs, how many times it's been verified, and which values it takes as parameters.
What it costs
Engagements are priced at three times the equivalent self-serve mission, on top of the normal price for that plan's composition. The multiplier is itemised on the receipt alongside the other factors, so you can always see what the service premium contributed.
An engagement freezes credits at the start and settles exactly like any other mission: charged in full only on a verified outcome, refunded in full on a platform-side failure, and partially recovered when work was genuinely delivered but could not be verified. Paying a premium does not change what happens when something fails.
Forward-Deployed Engineer is SnapFlow's agent swarm working an engagement end to end, not a human consultant assigned to your account. Everything it does is visible in the Flight Log, and every consequential change still stops for your approval.
SnapFlow