Rottawhite — AI Systems Studio

How do you know it is still right?

Most AI systems are measured once, on launch day, by the people who built them. Then a prompt gets edited, a model gets deprecated, a carrier redesigns a form — and nobody finds out that accuracy moved.

See it on your documents — $2,500 The audit trail

The problem

Silent degradation is the expensive kind

A traditional bug announces itself. An AI system that has quietly dropped from 94% to 81% field accuracy does not — it keeps returning confident answers, and the wrong ones look exactly like the right ones. In back-office insurance work that difference shows up months later, in an E&O conversation or an audit.

The only defence is to measure continuously against a fixed set of known-correct answers. That is all an eval harness is, and it is remarkable how many vendors ship without one.

The mechanism

What we actually build

A labelled test set from your documents

Built from your real submissions, policies, or loss runs — deliberately weighted toward the hard cases. The scanned fax, the manuscript form, the carrier whose layout changed last year.

Field-level scoring

Accuracy per field, not one number for the document. A system that nails the named insured and fumbles the deductible is not 90% right in any way that matters.

A gate on every change

Prompt edits, model swaps, and code changes are scored before they ship. If accuracy drops below the agreed threshold, the change does not go out.

Segmented reporting

Broken out by document type and source, so a weak format surfaces instead of being averaged away by the easy cases.

In practice

What this looks like on a run

An illustrative scoring run for a submission-intake system. The shape is what matters: per-field, per-segment, with a threshold that blocks the release.

FieldClean PDFScannedThreshold
Named insured99.1%96.4%95%pass
Effective date98.7%94.2%95%fail
Total insured value97.3%95.8%95%pass
Deductible96.9%93.1%95%fail

Two fields fail on scanned documents. That is not a reason to abandon the system — it is a reason to route scanned effective dates and deductibles to a human until the number comes up. Without the harness you would not know which two fields, or that scans were the problem at all.

Questions

What people ask

How big does a test set need to be?

Smaller than people expect. One to two hundred labelled examples, chosen to cover the document types and edge cases you actually receive, catches most regressions. A thousand random samples that are all the clean case is worth less than eighty deliberately awkward ones.

Who labels it?

We build the first version during the sprint and your team corrects it, because the person who does the work is the one who knows what the right answer is. After that it grows from the exceptions the system routes to humans.

What happens when a model is deprecated?

The replacement is scored against the same test set before it goes anywhere near production. A model upgrade becomes a measured decision with a number attached rather than a forced migration you hope went fine.

Is this not just unit testing?

It is the same instinct applied to a non-deterministic system. The difference is that the pass condition is statistical — accuracy above a threshold on a population — rather than a single assertion being true.

Related

The other half

Evals answer is it right. The audit trail answers how did it arrive at that. Regulated work needs both, which is why they ship together rather than as options.

Both are part of every production build, and the test set is yours to keep from the end of the sprint onward.

Next step

Ask a vendor for their accuracy number.

Then ask how they measured it, and when they last re-ran it. We would rather you asked us that question than took our word for anything.

Book a 30-min call The $2,500 sprint