July 29, 2026
About LabKey
AI is only as good as the data it runs on. In most labs right now, that data lives across 3 or 4 disconnected systems, exported to spreadsheets, copied into chat windows, stripped of the context and permissions it had a minute earlier. Ask an AI tool a question against that kind of data and you get a plausible-sounding answer. You don’t necessarily get a trustworthy one.
Most vendors are shipping chat interfaces onto whatever data model they already had, and most of those models were never built to meet the needs of large language models. What’s underneath the AI matters more than what it can do.
LabKey AI, announced today and now available, is LabKey’s answer to that problem: an intelligence layer built directly into the structured data model, permissioning framework, and audit infrastructure that already govern LabKey’s lab software, including LabKey LIMS, Biologics LIMS, Sample Manager, and SDMS. It lets scientists, lab managers, and IT teams query, understand, and act on lab data using plain language.
Before AI earns a place in a regulated lab, 4 things have to be true:
Results have to come from real, structured data. When you ask a question, you need an answer pulled from what’s true right now, with a clear line back to where it came from. Ask how many samples of a given volume and status are sitting in a freezer, and you should get a real query against live data and the exact count.
You have to be able to see the reasoning. An answer without a visible path to how it was reached asks you to take it on faith, and in a lab, that’s not good enough. You should be able to see logic behind how a question was answered, and verify that logic before you rely on what it returned.
Every interaction needs a record. Lab work needs to be auditable by default: who did what, when, and why. AI-assisted work has to meet that same bar. A system that can’t tell you who asked what and when doesn’t belong in a regulated workflow.
Permissions don’t get a shortcut. Whatever access controls already govern a lab’s data have to hold at every layer, including wherever AI touches it. A scientist with read-only access to a project gets answers from that project only, nothing more, and customer data is never used to train general AI models.
This is why LabKey’s products and AI capabilities are well suited for the rigors of the lab. The data model, the permission framework, and the audit trail that already govern work in the lab are the same ones LabKey AI runs on. Nothing new had to be invented to make it accountable, because the platform was already built to be accountable.
For bench scientists, that means a verifiable answer without writing a query. For IT, security, and QA teams, it means the same permissions and audit trail already in place, with no new risk surface, and the option to turn the capability off entirely for teams not yet ready to adopt it. For lab leadership, it means fewer bottlenecks waiting on a specialist to pull a report.
“AI is only as good as the data and rules beneath it,” said Michael Gersch, CEO at LabKey. “We have spent years building that foundation. LabKey AI is what that foundation makes possible.”
Any user, asking any question, in plain language, working from data that was already structured, permissioned, and audited before AI ever touched it. That’s the bar we built to, and it doesn’t move as the platform grows. In the next few months, expect LabKey AI’s capabilities to keep growing, changing how scientists report, analyze, and manage their data and workflows.
