Agents
Agents on both sides. Ours or yours.
Hospitals and labs run agents on their own data. Research teams run theirs. Biobase is the trusted layer where they work together.
Both sides
One trusted layer between them.
Hospitals and labs
Their agents, their data
- Specimen release agentChecks each request against your protocol and prepares the release for your approval.
- Coding agentSuggests CPT and ICD-10 codes from the final report. Your coders decide.
- Trial matching agentFinds open trials that fit the patients you diagnose.
- Slide QC agentChecks scanned slides for blur, folds and pen marks before they ship.
- Your own agentsBring an agent you built or bought. It works under the same checks.
Your systems or ours
Biobase
Governed exchange
- Consent
- De-ID
- Contract
- Audit
Screened before anything moves
Pharma, biotech and AI
Their agents, their models
- Cohort building agentTurns a protocol into a cohort request and searches partner sites for it.
- Sample procurement agentPlaces orders and tracks them against your study targets.
- Trial feasibility agentCounts eligible patients for a protocol, with a source for each number.
- Program matching agentWatches new supply and matches it to the programs you run.
- Your own agentsConnect an agent you built, on the model you chose. It works under the same checks.
Your environment or ours
Inside every institution
Our agents or yours. Any model.
Assemble
Our agents or yours, on Claude, GPT, Gemini or open models.
Context
Connected to Epic, lab, imaging and biobank systems.
Memory
Every request, decision and approval carries forward.
Collaboration
Agents work together across teams and across the network.
- Screened before it moves
- Your environment or ours
- A person approves
- Full audit trail
Pathology coding, with your coders in charge.
It reads each final report and suggests CPT and ICD-10 codes. It flags missed or wrong codes before claims go out.
- Your coders accept or change each suggestion.
- It reads reports from your LIS feed.
- It runs under a BAA with your lab.
Final report
Colon biopsies, three parts: A, B and C, each sent separately.
How a pilot runs
- 1
Look back
Compare past claims with what the reports support.
Agent + your team - 2
Watch live
Suggest codes on new cases, side by side.
Agent - 3
Work it in
Coders use it in their daily queue.
Your coders
An agent drafts. A person decides.
- 1
Drafts, not decisions
No agent output takes effect until a person accepts it.
- 2
Scoped
An agent works on one task, under your agreement.
- 3
Never to a payer
The coding agent suggests codes. It never submits a claim.
Questions about agents.
Does the coding agent submit claims?
No. It suggests codes. Your coders decide.
The agent never sends anything to a payer. It is a second reader for your coding team.
What does a pilot need from us?
Final reports and the codes you billed.
A look back on past cases comes first, so your team sees the differences before anything changes in the daily work.
Which LIS do you work with?
Tell us which one you run.
The agent reads reports from your LIS feed. We map your LIS data to one shape.
Is this part of the specimen program?
No. It is a separate agreement.
The coding agent runs under its own BAA. You can join one without the other.