Most of what a lab knows is written down. Almost none of it can be queried. Agents that read and write the notebook change that without changing how scientists work.
AI lab automation for protocols and ELNs means agents that turn free-text notebook entries into structured records, draft new protocols from what worked before, and flag where a run departed from its method. They work inside the electronic lab notebook and LIMS you already use, such as Benchling, LabArchives or a LIMS like LabWare, and every record they touch stays attributable to the scientist who signs it.
This is the software side of lab automation, not robotics. If you run liquid handlers or an automated platform, the agents can prepare and check the run files, but the hardware integration and its qualification are separate work, and we say so at the start.
Extraction is measured field by field against entries your scientists have labelled, and the hard cases are the ones we collect deliberately: shorthand, units written three ways, a value corrected in a later line. Protocol drafts are judged by the people who would run them. Where the notebook is in GxP scope, the system is designed to be validated, with versioned models and prompts and a complete audit trail; the validation itself, and the decision on which records are in scope, belong to your quality function.
Yes, through its API or export where one exists. Benchling, LabArchives and most modern ELNs expose entries, schemas and results programmatically, so agents can read entries and write structured fields or draft entries back. Older systems sometimes need an export route. The integration is usually most of the build, and we scope it on a sample of your real entries before quoting.
Part 11 applies to the records and the system, not to AI as such. The design requirements are attributable actions, a full audit trail, controlled access and validated software. We build the agent so it never signs and every change it proposes is logged and approved by a named person. Whether the result satisfies Part 11 or Annex 11 in your context is your quality function’s determination.
Most of the value here. The time a lab loses is largely in documentation: writing up runs, finding what was done before, reconciling methods with results. Agents that structure, search and draft notebook content help a fully manual lab as much as an automated one. Robotics integration is a separate project, which we scope separately.
What we built, what stage it is at, and a link to the live thing wherever there is one.
yourOS is an operating system with an AI at its heart. You talk to it like a person, out loud or typed. It makes your calls, plans your days and works on your projects with a crew of agents, on your own private server. Our own product, built from the first commit; the waitlist is open.
Website, backend and app assets for Sonora, a free AI sound-therapy app on iOS and Android published by Cognitive AI.
Brand, website, prototype, apps on five platforms, the AI backend and the algorithm-intelligence pipeline for Podswap, a social growth engine that runs without a human in the loop.
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Live at coherence.com: brand identity generated with AI agents we directed, the motion system and website for the group behind Sonora, AboutAI, WakeUp and MudShadows. Agent work remains in build.
Live at domainbooth.com for DomainBooth, a premium domain brokerage: a new animated mark, every page redesigned and rebuilt with designed mobile layouts, and a custom CMS for its sales record, featured names, press and enquiries.
Bring a sample of real entries. We will show what can be structured and what cannot.
30-minute technical call · no deck