Built by validation people, for regulated work.
Our founders have spent more than twenty years validating systems for Life Sciences manufacturers. dAInta is designed so that what it produces stands up to an audit.
How we use AI in a GxP setting
People approve, AI proposes
Risk ratings, matches, test protocols and reports are proposals until a qualified person approves them. Nothing enters the validated record automatically.
Every fact traces to its source
Each requirement, capability and finding links to the document, section and page it came from, and opens the original at that page.
Nothing invented
Every piece of text the AI extracts is checked word for word against the source section, and every proposed match must show its evidence. Anything that fails is rejected, not patched.
Same documents, same findings
Findings come from closed, versioned rule sets authored by validation experts, not free-form AI opinion, so the same documents give the same findings every time.
Audit trail and e-signatures
Every extraction, edit, override and approval is recorded with who, when and the value before and after. Approvals and test execution use electronic signatures designed to meet 21 CFR Part 11 and Annex 11.
Reproducible outputs
Every report records the data versions, template, rule sets and AI model it was built from, so it can be reconstructed exactly at audit.
The expectations dAInta is designed around
dAInta supports your compliance with these regulations and guidance. Compliance itself always depends on how a system is used and validated at your site.
| Regulation or guidance | Covers | How dAInta supports it |
|---|---|---|
| 21 CFR Part 11 | Electronic records and signatures | Electronic signatures for approval and execution, append-only audit trail, attributable actions, role-based access and multi-factor sign-in |
| EU GMP Annex 11 | Computerised systems | Requirement traceability, audit trail, controlled versions and risk-based testing |
| EU GMP Annex 15 | Qualification and validation | FAT, SAT, IQ and OQ planning, with earlier results leveraged into qualification where appropriate |
| GAMP® 5 (2nd ed.) | Risk-based approach to GxP systems | GAMP category set per system, lifecycle from URS to report, functional risk assessment drafted and checked for coverage before testing |
| FDA CSA guidance | Computer Software Assurance | Test effort allocated by risk across test phases and controls |
| ALCOA+ | Data integrity | Every data point attributable, original preserved, history kept in full |
| EU GMP Annex 22 (draft) | Artificial intelligence in GMP | Human review of all AI output, explainable findings, versioned and tested AI components |
We validate dAInta the way we expect you to validate it.
dAInta is developed and released under a documented quality process. Every AI component, from the rule sets to the prompts and the model in use, is versioned and tested against expert-reviewed reference documents before release. As a supplier, we will provide a validation pack to support your own assessment of the platform.
Your documents and data are stored in Ireland, encrypted in transit and at rest, and kept separate from every other customer. Sign-in uses multi-factor authentication. Full security and data handling details are shared with every early access customer, and we are happy to complete your supplier questionnaire or audit.
Join the early access programme
We are working with a small group of Life Sciences teams to shape the first commercial release. Early access customers get preferential terms and a direct line to the product team.