01A different kind of model for a different kind of work
Inpatient coding review means hundreds of small judgments per stay. Is the condition current? Is the stage documented? Is there MEAT (monitor, evaluate, assess, treat)? LLMs can make those calls, but at that volume they are slow, costly and hard to audit.
Jev is a System One model. Instead of writing prose, it answers structured questions, like multiple choice or yes/no, with calibrated probabilities. Deterministic rules turn those answers into results. There is no LLM anywhere in the pipeline, and Jev never even sees an ICD-10 code.
02The ICD-10 Code Reviewer
Our first agent on Jev checks a coder's draft codes against the whole chart, with evidence cited for every call.
- 1Pull the stayThe Kartha MCP FHIR Server provides the encounter, the draft codes, labs and every clinical note, including the discharge summary.
- 2Ask Jev small questionsHundreds per request: is it current, what stage or side, what caused it, is there MEAT, was it present on admission, is anything treated but not coded.
- 3Rules build the reportCoding rules turn Jev’s answers into findings. Each one cites the exact sentence it came from.
What the report surfaces
Nothing is auto-assigned. The coder stays in charge.
03One platform, both kinds of agents
- LLM-powered agentsChat and reasoning agents with the Agent SDK, 50 clinical skills and full receipts on every answer.
- Purpose-built modelsAgentic workflows on models like Jev, at millisecond speed and a fraction of the cost. Same MCP FHIR Server, same guardrails.
A word on PHI. The reviewer removes identifiers from note text before anything is sent to Jev, but redaction alone is not a compliance plan. For real patient data, plan for a BAA with , or run an open-source variant of Jev in your own cloud.
40 inpatient stays reviewed for 17 cents.
See the whole run in the demo, or talk to us about bringing the reviewer to your coding team.