A Deliberate Rep: Turning a Weak Status Update Into Proof of Judgment

Theory is cheap. Here's a deliberate rep, start to finish, on a task you probably do every week: a project status update for a skeptical VP. It takes about ten minutes — and watching where the capability actually gets built is the whole point.

The task, and the bar

Your project is slightly behind. You need a ~150-word update your VP reads in thirty seconds and can act on. Before you touch AI, set the bar — the three criteria a tough reviewer would use:

1. Leads with the decision or ask, not a list of activity.
2. States the risk honestly, with a mitigation.
3. Survives one skeptical follow-up — "so what do you actually need from me?"

Writing the rubric first is the move most people skip. It's also the move that makes everything after it work.

The vending-machine version

Prompt: "Write a status update for my project." You get something fluent and useless:

This week the team made strong progress across several workstreams and remained committed to our goals. We encountered some challenges but are actively working to address them and remain optimistic about delivery.

It reads fine. Now score it against the rubric: leads with vague activity (fail), buries the risk in "some challenges" (fail), no ask at all (fail). 0 / 3. This is exactly what most people ship — fluent, confident, and empty.

The rep

Now do it deliberately. Give the model the real context: a two-week slip caused by a dependency, and the fact that you need a scope decision. Iterate twice. You land much closer — but read it against the rubric and two gaps remain: the risk is softened to "a minor delay," and the ask is implied rather than stated.

Score, then close the gap yourself

This is where the capability lives. The model produced a good-enough draft; your judgment finishes it:

• "Minor delay" becomes "a two-week slip on the API dependency."
• The implied ask becomes explicit: "I need a decision on cutting scope vs. moving the date by Friday."

The final artifact:

Status — Project Atlas (week of the 16th). We're tracking two weeks behind on the API dependency; root cause is a vendor delay, now escalated. Everything else is on plan. Decision needed by Friday: cut the export feature to hold the date, or move the launch two weeks. My recommendation is to cut scope — details attached.

Thirty seconds to read. Impossible to misread. One clear ask.

What this rep proves

Notice what actually built the capability. It wasn't the AI draft — the model wrote both the useless version and the good one. It was the rubric (it caught all three failures) and your judgment (it closed the gap the model couldn't see — the honesty about the slip, the explicit ask).

That final update, with your fingerprints on the risk framing and the decision, is a portfolio artifact: proof you can turn a fluent draft into a decision-ready one. The model is a commodity. The judgment that did the closing is not.

This is one rep. The UofAi method is built to give you dozens of them, each one evidence. Start free.