Case study · AI adoption

AI-supported R&D
evidence workflow.

The organization needed a defensible view of R&D tax-credit qualification before a leadership meeting in three business days. The work required becoming familiar enough with the relevant criteria to evaluate work credibly, then reviewing a large body of evidence under tight time pressure.

My role

What I was responsible for.

Release Train Engineer building the evaluation framework and AI-supported qualification workflow alongside cross-team planning and delivery governance across roughly 10 teams.

Approach

How the work moved forward.

01Built an evaluation framework before writing prompts or scaling tool use.
02Used AI to query and assess work against approved sources, with named human reviewers for every inclusion decision.
03Produced a traceability spreadsheet so leaders could defend results and question any line back to source material.
04Kept decision authority with leadership: the workflow accelerated analysis, not accountability.

Outcome

What changed.

Leaders entered the meeting with a reviewed qualification view, traceable evidence, and clear items for judgment. Work estimated at roughly four weeks manually was completed in three business days, with confidence in the decisions still resting with the people responsible for them.

Sample deliverables