← Products/Locus
LocusAI for Science · Biomedicine
Let AI truly
take part in research.
We are building an AI research system that can design, run and analyze experiments on its own. The first step is its brain.
01Roadmap
First the brain. Then the hands.
- We are hereSTEP 01
Research brainLocus
AI planning, real computation, checks and retained results.
Engineering Alpha · In use - STEP 02
Connect the instruments
Turn laboratory instruments into tools that AI can call.
In development - STEP 03
Autonomous laboratory
Give it a goal: AI schedules instruments around experimental cycles, analyzes data and decides the next step.
Vision
02What it is · About Locus
AI that does research,
and keeps the evidence.
Locus is an AI research system for life science. You pose a research question; it plans the analysis, runs it in a real computing environment, reads the results and decides what comes next. Every attempt, piece of code, dataset and check is recorded. It distinguishes a completed computation from a supported conclusion, so each result can be traced, reviewed and used in further research.
Research projects
Cancer cell line RNA-seqDifferential expression · 4 steps Single-cell atlasCell-type annotation · Draft Pathway enrichment reviewArchivedResearch questionWhich genes are significantly upregulated in the treatment group compared with the control? Which pathways are they concentrated in?
- Read data and check qualityDone
- Alignment and quantificationDone
- Differential expression analysisDone
- Pathway enrichmentComputing
Output · Differential expression volcano plot
- The model answered
- Computation finished
- Files were written
- Numerical checks passed
- Interpretation holds · Awaiting researcher review
Illustrative interface
03How it works
A research cycle that keeps going.
- Pose a questionAsk
- Plan with AIPlan
- Run computationsCompute
- Observe feedbackObserve
- Check the resultsCheck
- Record resultsRecord
- Continue researchContinue
The model plans, a real environment computes, and the findings of checks are recorded separately. AI cannot declare its own work “verified.”
Five levels of completionRecorded separately and independently
- 01The model answeredAnsweredRecorded
- 02Computation finishedComputedRecorded
- 03Files were writtenFiles writtenRecorded
- 04Numerical checks passedValues checkedRecorded
- 05Scientific interpretation holdsInterpretation holdsAwaiting researcher review
04Why Locus · Core values
Every step of research counts.
- 01Research accumulatesResearch that accumulates
Every question, attempt and result is retained. Later success does not silently overwrite earlier failure. Previous results can be checked and reused in the next study.
- 02Every step in viewEvery step in view
Code, logs, data sources and the scope of checks are open for inspection. You can see where a result came from, what has been checked and what has not.
- 03No lock-inNo lock-in
The model, agent core and computing environment are separate. As technology changes, the research assets you have built remain in your hands.
Compute locally. Control what is sent out.
05Progress
Where we are now.
- Stage
- Engineering Alpha Engineering Alpha
- Verified so far
- Complete RNA analysis workflow Compute, view, export and reopen
- Selected single-cell analyses
- In progress
- Stability of autonomous analysis
- Broader coverage of analysis methods
06Next
Concept film · Not live footage: the direction we are working toward
Instruments as tools
AI can call each instrument as needed, with the experimental plan and parameters.
Advance with the experimental cycle
The next step starts when it is due, without someone standing by to press a button.
Human confirmation for critical consumables
Steps involving critical consumables always require researcher confirmation before execution.
07Get involved
Build it with us.
Apply for an Alpha trial
Try it with your real data and help us refine Locus.
Apply for a trialBecome a partner lab
Connect instruments with us and jointly validate autonomous experimental workflows.
Apply to partnerInvestment and partnerships
Explore the roadmap, current progress and ways to collaborate.
Start a conversation