Small agents, aimed at real delivery pain.

An internal prompt-engineering assistant built on a nine-part framework, benchmarked across Claude, ChatGPT, Perplexity and Gemini so the team gets consistent output whichever model they reach for.

Turns raw call transcripts into structured Confluence documentation — summary, action items, risks, decisions and open questions — written the way my delivery team already reads them.

A delivery-risk forecaster. Reads burn rate, ticket velocity and sentiment from meeting notes to flag programs likely to slip before the weekly status meeting does.

Drafts a first-pass SOW from a sales call transcript: assumptions, exclusions, phased milestones and an estimate range with the reasoning attached.

A client-health dashboard that scores accounts on delivery signals and relationship signals side by side, so renewals stop being a gut call.

A Slack agent that assembles the weekly client status report from tickets, commits and call notes, then waits for a human to approve it.