01 — Work Emerging Technology Vendor Evaluation
AI-Assisted Vendor Diligence Brief
AI-assisted diligence brief
Open the verification workbench
AI-assisted diligence — representative
Evidence gaps identified
- Regulatory readiness — unclear from materials provided
- Implementation support — timeline not substantiated
Follow-up questions developed
- Request documentation on regulatory review status
- Confirm implementation support model and timeline
Reviewed by Alicia — refined before use in diligence
1. How the brief was built
Built early in diligence from vendor materials, evaluation criteria, and prior findings — a starting point for the team’s thinking, never a finished conclusion. Every claim in the draft still had to survive a check against the source material before it reached a stakeholder.
2. Catching an unsupported claim
Representative example — one evidence gap, traced from AI draft to refined brief
“Implementation typically completes within 4–6 weeks, based on the vendor’s standard onboarding process.”
⚠ No source in the materials provided named a timeframe — the draft turned a general description into a specific number.
“Implementation support — timeline not substantiated in materials provided.”
Follow-up question added: “Request a documented implementation timeline, with milestones and a named point of contact.”
This is the pattern the review step existed to catch: AI output that reads as specific and confident even when the underlying evidence is general or missing. Finding the gap — and turning it into a sharper question instead of a false data point — was manual work, done before anything reached the scorecard or a stakeholder.
3. Evidence gaps & the AI/human split
| Evidence gap | What the AI draft said | Follow-up question developed |
|---|---|---|
| Implementation support | Stated a specific onboarding timeline | Request a documented implementation timeline and milestones |
| Regulatory readiness | Stated regulatory review was underway without a cited source | Request documentation on regulatory review status |
What the AI did
- Synthesized vendor materials, criteria, and prior findings into a first draft
- Surfaced candidate evidence gaps and draft language for follow-up questions
- Produced language that read as confident regardless of how strong the underlying evidence was
What Alicia did
- Checked every claim in the draft against the original source material
- Removed or downgraded claims the materials didn’t actually support
- Decided which evidence gaps mattered enough to become follow-up questions
- Kept the AI output out of the scorecard and the final recommendation entirely
Problem or decision supported
Synthesizing vendor materials, evaluation criteria, and prior findings by hand was slow and made it easy to lose track of where the evidence was thin. The brief needed to surface those gaps quickly without being mistaken for a finished conclusion.
My contribution
Directed the AI synthesis of vendor materials, evaluation criteria, and known evidence into a structured brief — then reviewed and refined the output, removed weak or unsupported conclusions, and used the strongest evidence gaps to develop more detailed vendor follow-up questions.
How it was used
Generated early in diligence, then reviewed and refined before use — never treated as a recommendation or source of truth. It supported vendor follow-up questions, technical deep dives, regulatory review, implementation diligence, scorecard updates, unresolved-question tracking, and executive decision preparation.
Outcome or decision enabled
Gave the team a faster, more consistent way to spot evidence gaps and turn them into sharper follow-up questions, without letting AI output stand in for stakeholder judgment or a final recommendation.
Tools and methods
AI-assisted synthesis, reviewed and refined by hand before use in diligence
Intended audience
Cross-functional stakeholders participating in vendor evaluation
Confidentiality Representative reconstruction — real vendor names, findings, and internal conclusions are not reproduced.