An AI-native research execution platform for GLP preclinical studies.
Capture and check study work at the point of action — at the bench, before it becomes a permanent record.
The problem
Bench work gets recorded twice: once by hand, once into a database. Every transcription is an opportunity for error, and in a GLP environment correcting one is not an edit. It is a documented change, a reason for change, and a review.
Protocol deviations — a blood draw outside its window, a formulation past its stability limit — typically surface weeks later, during retrospective Quality Assurance review. By then the study is finished.
SEND dataset compilation runs on manual spreadsheet mapping and terminology translation, weeks of it, after the study is already done. Datasets that do not conform put a submission at risk of Refuse-to-File.
The approach
Hands-free capture at the bench, in the operator's own words, structured into protocol-bound fields as the work happens — preserving attribution and contemporaneity rather than reconstructing them afterward.
Deviation detection at the moment of entry. The event is always recorded as it occurred; the system annotates it with the deviation and the required reason for change. It flags and routes. It never blocks the record — blocking what actually happened would itself violate contemporaneity.
Audit-trail triage that classifies materiality across the complete change history, surfacing the consequential minority of events a reviewer needs to read. The full log is always preserved and available.
Built for the record
Built for 21 CFR Part 11 environments and designed for GLP workflows, with ALCOA+ at the point of action.
Every agent action is deterministically bound to an append-only audit record. The system may assert only what it can cite to the governing protocol or standard operating procedure. Models are frozen in production under formal change control, with complete decision logs.
No generated output enters the regulated record without human review. The Study Director signs. Always.
Who it is for
Small and mid-sized contract research organizations, academic toxicology centers, and government testing laboratories — the long tail that still runs on paper, spreadsheets, and disconnected instruments because the enterprise suites cost too much and take too long to stand up.
Contact
Boots Sentinel is early and in active development. If you run GLP studies and this maps to problems you have, we would like to hear from you.
Or write directly: damara@bootsagentai.com