Three records that stopped agreeing
Every module has three records of what it's supposed to do: the user stories, the prototype and the spec. By handoff they rarely agree. The first full review of one OMNE module found about 175 issues between them.
175 issues in one module
That was vendor invoicing, in September 2026. Its first review before handoff compared the stories, the prototype and the spec, and found about 175 issues. By the latest round on 23 September it was down to about 30, mostly prototype data that didn't add up.
This isn't unusual. Stories get written early, and the prototype and spec keep changing through every review. If nobody checks all three against each other, they drift.
What drift looks like
Story, no screen
A story asks for something the prototype can't do.
Screen, no story
The prototype does something no story asks for, so nobody will test it.
Spec vs story
The spec and the story give two different rules for the same thing.
Data that disagrees
Numbers in the prototype contradict each other. That often hides wrong logic.
The numbers
Issues found in the first and the latest review round of one module, September 2026.
Findings in one review that were the prototype misbehaving. The stories were fine.
Issues found in a review of one module's 52 user stories.
Stories read in one module's handoff review. It came back with 33 findings and 11 open decisions.
Source: the Omne decision wiki, counted on 16 November 2026. The counts grow every night.
Not every finding is real
In the first vendor invoicing review, about 26 of 40 findings turned out to be the prototype misbehaving, with the story itself fine. They still needed fixing, just by someone else.
It goes the other way too. On another module, about 30 findings over three rounds got waved off as "only data issues". They were real errors. Bad data in a prototype usually means bad logic somewhere.
Sorting the real problems from the noise is where most of the review time goes.
What it changes for engineering
Every gap you catch before handoff is one the developers don't build. The team's goal is that no business analysis or design issue reaches engineering or QA.
By late September 2026, after the team made answering engineers' questions a priority, engineers had gone almost a month without a fair complaint about unanswered questions.
If your developers keep finding gaps after handoff, or your stories and designs say different things, tell me what's stuck. The AI case study shows where reviews sit in the team at OMNE.
More of the work
Tell me what's stuck
Send a few lines about your product, your team and the problem. If I'm not the right person for it, I'll say so.