Five applications, one lab, one record underneath.

One workspace spans samples, instruments, freezers, robots, and rooms, instead of one system per island each with its own login and its own version of the truth. Provisioned at sign-up, with your team invited into it under their own roles.

Facility and welfare data is some of the most sensitive data a lab holds. Run in your own cloud tenant and it never leaves infrastructure you control — with the same audit trail, access model, and compliance behaviour as the managed tier.

A new alert threshold, a rebalanced robot schedule, a changed escalation path — drafted, reviewed, committed, and reversible. Operational change stops being a deployment and becomes something a lab manager can make and a QA reviewer can read.

The freezer, the incubator, the cage rack, the robotic arm, the LIMS, and the badge reader all attach through the same two node types — polled or pushed. OPC UA and SCADA, which most lab automation and building systems speak natively, are roadmap; today those systems connect through their REST, file, or broker interfaces.

Cold gives you a year of temperature history for an excursion investigation. Warm answers which units are drifting right now. Hot pushes a live feed to the operations screen, and a change feed fires whenever the workspace configuration itself changes.

Robotics and instrument vendors ship islands, and a new node type is how an unintegrated arm, reader, or cage sensor becomes a first-class participant in the same flow as everything else. Built by our team on the shipped node model today; self-service authoring is roadmap.

Ask which freezers drifted last night, which robot runs stalled, or which racks stopped reporting — and get an answer grounded in your actual workspace. The assistant drafts new monitors, queries, and dashboards as proposals for a person to approve.

Any MCP-native model can read the lab's live state, run queries across it, and propose new monitoring, but cannot enact anything. The absence of an accept tool is the architecture, not a policy — which is what makes agents usable in a regulated facility at all.

A vivarium technician, a facilities engineer, a study director, and an external auditor each need a different lab. Named access rights bundled into custom configurations give each of them exactly their own, and every assignment is itself audited.

The audit machinery is already load-bearing for sample custody: attributable, timestamped, append-only, each event tagged with the ALCOA+ principle it satisfies. Monitoring runs on that same spine, so an excursion and its acknowledgement are evidence rather than a side channel.

Planned: promote a monitoring configuration or workflow change from a testing environment through staging into production on your own change-control steps — so a change to a facility's alerting is as controlled as a change to a study.

Audit-trail export is live today. Broader validation-data downloads and custom compliance reports and dashboards over operational and environmental data are planned.