ARIA - AI-powered program risk monitoring

ARIA is an AI-powered risk monitoring tool I built and deployed on a large enterprise modernization engagement. It watches a program's RAID log, risks, assumptions, issues, and dependencies, and automatically surfaces what needs attention, so the things that usually slip through a spreadsheet get flagged before they become problems.

The Problem

On large programs, risks and issues live in sprawling RAID logs that nobody has time to read closely. Important items get buried, dependencies go unnoticed, and problems surface too late. I was running program operations across a big engagement and kept seeing the same failure: the information existed, but no one could stay on top of all of it.

What I Built

ARIA, an AI layer over the RAID log that reads the entries the way a sharp program manager would and flags what actually matters, rising risks, stale items, dependencies about to bite. Instead of someone manually combing a spreadsheet, the tool does the first pass and points attention where it's needed.

How It Works

Built in React, deployed on Vercel, and powered by the Claude API. The AI interprets unstructured RAID entries and turns them into prioritized signal. Nobody asked me to build it, the work needed it, so I did, which is the pattern across most of what I make.

Outcome

What started as a tool for my own program spread to other teams on the engagement, eventually supporting work across nearly twenty teams and hundreds of people. It turned a passive log into an active monitor. More than the tool itself, it's proof of how I operate: I don't file a ticket and wait for engineering, I build the fix myself, on the Claude API, and put it to work.

Closing Thoughts

ARIA is the clearest example of what I mean by AI-native. The problem was a program-management problem, but the solution was software, and I could build it because I don't wait for permission to solve things.