
Understand the business, choose the stack, design the flow — access confirmed, interviews done, and the decision made about which one problem is worth the build.
The working thing, on their real data — with the people who’ll use it watching it take shape and correcting it early.
Demo, deploy, document, train, hand off — the part that decides whether it’s still running in March. (Classes 14 and 15 are entirely about doing this right.)
Enterprise software was bought through 9–12 months of evaluations, committees and proposals — claims about what the software would do.
Palantir — the company that named forward-deployed engineering — ran “bootcamps” for its AI platform (AIP): five days building a real workflow on the customer’s real data, before any contract. Demo Friday, decide Friday.
organizations in a single 2023 quarter — and ~75% of thoseconverted to contracts. One quarter, one company; not a lifetime win rate.
D0No access, no startData access confirmed before day one — or day one doesn’t happen.
D1MapInterviews and a systems inventory — who does what, where data lives.
D2DiveIngest real data; shadow the operators doing the real work.
D3DecideValue × feasibility — choose ONE demo. Everything else waits.
D4BuildBuild it with operators validating as it takes shape.
D5Demo + closeShow the working thing on their data. Friday is a decision event.
A business applies — the assessment is the application (deadline Sep 18, decisions week of Sep 28). Their answers become your first briefing document.
Five partners, one team each — weekly visits, weeks 9–14. This is Map and Dive at semester cadence: interviews, shadowing, real data.
Audit, spec, or working build — one primary deliverable, documented, trained, handed over. Demoed at the week-14 presentation. (Why only one? Class 13.)
Connect.AI
Checking access…