Scientist-led laboratory intelligence for connected, traceable omics workflows.

CLAIR
ConceptApproachWhy nowMissionWhoContact
Pre-launch laboratory intelligenceScientist-led prototypeSan Francisco / clair.bio
A pre-launch laboratory intelligence concept

A clean read on the biology you're changing.

A prototype for putting experimental design, lab workflows, analysis, and interpretation on one traceable, scientist-led operating layer.

Explore the conceptContact
Pre-launch
concept and prototype development
Scientist-led
expert judgment defines the gates
Traceable
provenance is designed in
01The bottleneck

The judgment that makes an assay work rarely reaches the protocol.

It’s tacit. Execution and analysis are split across teams, losing signal at every seam.

02The concept

One experimental thread, from question to interpretation.

Infrastructure connecting experimental design, lab workflows, analysis, and interpretation—with scientific judgment and provenance explicit.

Research contexts:

Low-input CUT&Tag / TIP-seqLow-input RNA-seq (Smart-seq3 / FLASH-seq)ATAC-seqMultiomeWGSDNA-seq+ methods with no kit
01
Experiment intent
objective + controls
02
Workflow design
methods + checkpoints
03
Execution context
instruments + samples
04
Analysis
planned comparisons
05
Interpretation
evidence + uncertainty

A model for a connected workflow—not a current sample-processing service.

03The design goal

Reduce handoffs without removing scientific judgment.

A shared operating layer should make workflows easier to coordinate, inspect, and learn from. Cost, speed, and quality targets remain to be tested.

Connected contextBounded automationExplicit provenanceInspectable decisions
04The intelligence layer

Well-documented experiments should make the system easier to inspect and improve.

Methods, QC decisions, and provenance become reusable structured evidence. How well that compounds is still under evaluation.

05The principle

Automate the repetition. Keep scientists responsible for the science.

Bounded, inspectable support for expert work—not the removal of scientific judgment.

06Why now
01
Agents can help coordinate workflows
Current models can assist with planning and interpretation inside explicit bounds.
02
Experts can encode checkpoints
Tacit craft can become inspectable methods, controls, and QC gates.
03
Trust must be earned
Prototype evidence, auditability, and human approval are prerequisites—not assumptions.
07The mission
Explore a less fragmented way to do biology.

Shared experimental context, so scientists coordinate more of a workflow without obscuring judgment, uncertainty, or provenance.

08Who’s building it

Clair is built by a bench scientist.

Di Hu is a full-stack biologist: wet lab, automation, computation, ML. The judgment described here is judgment she holds—which is why it can be encoded rather than translated.

01
Bench credentials
Oxford DPhil, Clarendon Scholar. Joint first author in Nature Communications; eleven peer-reviewed publications across epigenomics, single-cell neurogenomics, and developmental biology.
02
Automation in practice
Hamilton, Opentrons, Tecan, PyLabRobot—from low-input epigenome methods to high-throughput biochemical and cellular assays, and the automated systems that run them.
03
Built in the open
Cell sorter support contributed upstream to PyLabRobot (PR #1156, in review), alongside public simulation-first robotics, auditable MCP tooling, and workflow benchmarks.

The work is public: github.com/di-omics · PyLabRobot #1156 · reinhaudt.com.

→Contact

Follow Clair as it develops.

Clair is pre-launch. Get in touch about the concept or its research direction.

Please do not submit confidential information, proprietary project details, PHI, patient data, or other regulated data.

Prefer email? Write to di.autonomouslab@gmail.com.

Read the state you're changing.
Contact
Clair / di.autonomouslab@gmail.com / San Francisco / Pre-launch laboratory intelligence
Built by Di Hu · reinhaudt.com · github.com/di-omics