batteriesincluded.com · Questions & Answers

Should we ask a forward-deployed AI engineer for a fixed-price build before they have watched how our business actually works?

Ask for a bounded discovery engagement first, then price a defined build. A fixed price is useful when the outcome, dependencies, and acceptance criteria are clear. Before an engineer has observed the work, it can simply hide assumptions that later become change orders.

Forward-deployed engineering should bring the builder close enough to the operation to understand exceptions, not just the workflow leadership describes. Have the engineer follow a real job from request to completion, inspect the systems involved, and identify where people make judgment calls.

Keep discovery practical and low on red tape. Its deliverables should include:

- A map of the workflow and its consequential exceptions.
- A narrow proposed outcome, with explicit exclusions.
- Evidence that the required integrations are feasible.
- Acceptance criteria tied to business performance.
- A build estimate that identifies remaining uncertainty.

Keep EOS as the operational backbone. The discovery findings should inform the leadership team's existing priorities and Issues List, not create a competing transformation agenda.

For a Batteries Included build using Lovable or Replit, a working screen can help test understanding, but it is not proof that integration or operational risks are resolved. Ask the engineer to investigate the riskiest dependency before polishing the interface.

The recommendation is to buy reduced uncertainty before buying the full implementation. If discovery shows that a small change to an existing system would solve the problem, that should count as a useful outcome.

Category: Forward-Deployed Engineering

← All questions