The Scientific Harness
Harness the compounding power of every experiment.
.png)

The missing foundation
The leap to closed-loop lab automation that can meaningfully accelerate discovery cannot happen without first structuring the knowledge that underlies it. That foundation does not yet exist. We are building it.
CORTEX transforms today’s labs into the labs of the future
The labs of the future will be increasingly autonomous, giving scientists greater leverage to pursue discovery. CORTEX connects today’s laboratories to that future by turning accumulated knowledge into a living system that enables every experiment to build on what came before.

Modern biology is drowning in fragmentation
The laboratory is not legible to AI
AI can retrieve the data, but without its scientific context, it cannot reason over it effectively.

From fragmented records to scientific state
CORTEX connects the systems a laboratory already uses, resolves the relationships between them, and continuously builds the scientific context that AI can reason over.

Connect
CORTEX connects to ELNs, LIMS, instruments, files, archives, and other sources without requiring the laboratory to replace its existing infrastructure.

Resolve
CORTEX resolves entities, provenance, experiments, results, and decisions into a connected scientific representation.

Operate
CORTEX makes that scientific state usable by AI and captures verified outcomes back into the evolving state.

Everything an AI system needs to operate inside a real laboratory
CORTEX maintains scientific state, assembles task-specific context, governs interaction with the laboratory, and captures what happens next.
State
The persistent scientific record of the laboratory: entities, evidence, relationships, history, and judgment.
Context
Task-specific context assembly. CORTEX selects and assembles the scientific context needed for a particular question, reasoning task, or action.
Interfaces
The mechanisms for retrieving knowledge, invoking analysis, and interacting with external systems: memory and knowledge retrieval, analytical tools, software, instruments, and external systems.
Control
Scientific control logic that accounts for the laboratory’s physical, procedural, and implicit constraints, so the agent operates within the actual boundaries of the scientific environment.
Security
Encryption, permissions, and auditability across data and actions. A boundary around the whole harness.
Loop
Every verified outcome updates the scientific state, compounding knowledge and creating feedback for what comes next.
The same model performs better with CORTEX.
A controlled ablation using the same answer model with and without CORTEX.
CORTEX retrieval against the common approach
Context precision and faithfulness of retrieved context, scored with RAGAS on PubMedQA.
CORTEX in practice
Biotech archive
1.8M files becomes 8 relevant records.
~73,000 searchable files reduced to 8 relevant records for a specific scientific question.
“This is 50× better than Box AI. It’s like night and day.”
Read case study →
Research institute
50 TB of science becomes one connected map.
Millions of files and multiple systems reconstructed into a connected research record.
“It’s not a search function, it’s actually a mapping function.”
Read case study →
Preclinical program
3 studies becomes one connected research record.
Three historical studies connected across animals, observations, endpoints, and source-level provenance.
“The team has run six different data sets. Which one is reproducible? Which one’s believable? Where did that data come from?”
Read case study →

