AI Opportunity Triage: Scored, Drafted, Human-Approved

The problem. A busy pipeline gets hundreds of new opportunities a week. Most are a poor fit, and the good ones go cold while someone reads through the noise.

What I built. An AI triage-and-drafting layer that sits in front of the team:

Fit check: each opportunity is scored against real, verified capabilities. Anything generic or off-scope is filtered out.

Drafting: the AI writes a first response in the brand's voice, from the opportunity's actual text, never from a template.

Guardrails: duplicate messages and unbacked promises are blocked before anything can go out.

Human checkpoint: price, timeline or any commitment always waits for a person to approve.

How it stays trustworthy. Every change to the decision logic is replayed against 60 real historical decisions before it ships, the same way you would regression-test software before a release. Loosening a rule without that test is how AI systems quietly start making bad calls.

What clients get from this experience: automations that are fast and safe to leave running, with clear limits on what the AI may decide on its own.