ARTICLES · DUE DILIGENCE, ON US
Read the guides before you commit.
Decision guides and engineering notes on choosing, scoping, building and running AI systems that keep working in daily operations, long after launch.
- Read guides to choosing and building useful AI systems.
- Explore the engineering work behind reliable operation.
- Test a proposal before you commit budget or people to it.
Strategy
You Are Approving A Build Nobody Has Checked
The proposal has a number, a timeline and a team who believe in it. Most AI projects are lost in that room, because nobody independent has checked the decision it changes, the data it runs on, or the systems it must plug into.
Start With The Call That Can Say No
The first step you take should be the one whose best possible result is that you spend nothing else. Here is what a free first call owes you, when a written plan is worth paying for, and what that written plan has to contain.
Your Delivery Metrics Look Healthy While Risk Grows
Your DORA metrics, deployment frequency and recovery time, look healthy while teams quietly lose their hold on the systems they ship. AI makes this drift faster and harder to spot, so pair delivery metrics with comprehension.
When to Buy Delivery and When to Buy Ownership
A fixed-price build insures you against the system not existing; a monthly engagement insures against it existing with nobody responsible. Here is how to tell which failure you are exposed to, and what each contract covers.
A Spreadsheet Is A Control Until Someone Asks
Remediation fails on evidence, not process. The control usually exists and someone really performs it; what is missing is proof that it ran, in order, by the right person, before the period closed. Make the step record itself.
An AI Reviewer Without A Rubric Just Flatters You
A reviewer model without a rubric returns fluent, forgettable praise. GreatFeedback.ai, built and run by ML LABS, scores product feedback against named criteria and rates the reviewer itself, because structure is the product.
Measure AI Impact By Decisions, Not Accuracy
Model dashboards show accuracy and uptime while the business sees nothing change. This article shows where AI value actually escapes, and how to measure the two levels that decide it: outcomes and business impact.
Why The Plan And The Work Drift Apart
When documents and tasks disagree, someone's job becomes reconciling the two. Making the work and its requirements one connected record, with one type of node for both, keeps people and AI agents reading the same state.
Is Your Data Good Enough To Start Building?
You have a workflow worth fixing and no way to tell whether your data can carry it. Build-ready data is accessible, representative and stable enough for one workflow. This is the lowest bar it has to clear before a real build starts.
Who Signed Off, And What The Record Shows
After an incident, you need to know who approved the action and what happened. A mail thread and somebody's recollection of a call are not a control. Build those answers into the record before adding AI to a process.