Case studies
Working systems, shown with their safeguards and limits.
Start with Smart Inbox, our evaluated in-house case study. The guidance assistant and bookkeeper are build previews; their full case studies and evaluations will follow. These are BrisAI systems, and no client outcomes are claimed.
Proving an AI inbox sorter before a client relies on it
In-house evaluationOperations and admin158 of the 165 messages this system filed automatically were filed correctly, on a 250-item held-out set it was not designed from, with none of the seven errors high-severity. It is the fourth version, and the fourth pre-registered evaluation. Three of the four found something that had to be fixed, and all four are published — including the one that proved our own prediction wrong.Both readiness gates met · 250-item held-out evaluation
Answering business questions from official guidance
Working in-house systemKnowledge and supportAnswers questions from an approved set of official guidance and shows the page each answer came from — and declines when the material does not cover the question.Working demonstration · formal evaluation pending
Turning invoice attachments into a reviewable register
Working in-house systemFinance operationsFiles an incoming invoice in the right place and prepares a register row from it, with the extracted figures sitting next to the original document so a person can check them.Working system · formal evaluation pending
Reading these pages
What the evidence labels mean.
Each page carries one evidence status. The label describes exactly what exists behind it today, and never implies a client outcome.
- In-house evaluation
- BrisAI built the system and tested it on a declared evaluation set or controlled process. No client outcome is implied.
- Working in-house system
- A working BrisAI system available for demonstration, with safeguards and limitations documented, but no formal evaluation result yet.
Your process
The interesting question is what your version of this looks like.
If one of these resembles work your team does by hand, the useful next conversation is about your volumes, your exceptions and what has to stay under human control.