Topology
Instruments, deck positions, plate hotels, reachable transfer points, transport arms, and physical hand-off edges.
labspine.ai
The canonical control model for robot-powered biology.
Design a heterogeneous workcell, compile it into real instrument protocols, validate against simulation, and read the logs back when the pipeline breaks.
Labs do not fail in one place. A plate can run correctly, disappear at the hand-off, land under the wrong sample ID, and surface later as bad bioinformatics. LabSpine follows the sample through the whole chain.
canonical cell model
Instruments, deck positions, plate hotels, reachable transfer points, transport arms, and physical hand-off edges.
Operation ranges, labware support, heads, consumables, throughput, I/O contracts, and vendor-specific limits.
The assay as a dependency graph with timing constraints, sample identity, labware lifecycle, and resource needs.
validate, repair, diagnose
LabSpine treats LLMs as proposers, not authorities. Every design becomes a structured patch to the model, then passes through catalog checks, solver constraints, simulation, and human gates.
interactive proof path
diagnostician verdict
Sample S003 completed the cell run but never produced a matching LIMS hand-off event.
Ready
JavaScript is disabled, so the scenario buttons cannot switch cases. The table below shows the default silent hand-off drop.
| Sample | Cell | Hand-off | LIMS | Bioinformatics | Landed as |
|---|---|---|---|---|---|
| S001 | OK | OK | OK | OK | S001 |
| S002 | OK | OK | OK | OK | S002 |
| S003 | OK | FAIL | FAIL | FAIL | - |
api findings
Run a live scenario to view generated artifacts.
labspine.ai
LabSpine is a control-plane product for automated labs: one model that supports design, validation, compiler output, and operational diagnostics.
A thin vertical slice already compiles real Opentrons protocol code, validates it, repairs infeasible designs, and localizes sample failures across the pipeline.