How we work

Evidence over adjectives.

We'd rather show you what our software does, and say plainly what it doesn't do yet.

Built and tested internally on synthetic material, being validated now.

Principles

Three habits we keep.

01

Every claim maps to a test

If we say the software does something, there is a test that shows it. Internal tests are called internal tests, never customer results.

02

We publish what's open

We list what is still in development, and we don't name customers or quote numbers we can't back.

03

Your systems stay in charge

We work beside the systems you already run and respect the access rights you already manage.

Extend, don't replace

Safe, incremental, and reversible.

  • Your hospital information system remains the system of record.
  • New capability arrives one site, programme or workflow at a time, with a way back.
  • Modules switch on only when they are ready for that site, never just because they are installed.
  • Each site assesses the intended use of each clinical capability before it is released.

The boundary

What never changes.

Each site confirms its own privacy and breach procedures.

Patient data stays within your hospital-controlled boundary.LevCore runs on-premise, one hub per hospital.
Your patients' records are never used to train a model.A fixed rule of the design, not a setting.
Each reviewed answer carries signed, checkable evidence.Who approved it, against which source version, and when it was withdrawn.

Where we are

Built and tested internally on synthetic material, being validated now.

Built and tested internally
  • Approved answers bound to the exact source version, withdrawn when the source changes
  • Search over the sources each person may see, by keyword
  • Signed, checkable evidence for each reviewed answer
  • Other software agents can find, fetch and recheck approved answers
  • PulseEdge offline workflow, on synthetic records
In development
  • Patient-scoped evidence retrieval for clinician review
  • Text recognition for scanned records at scale
  • Live connectors to document management systems
  • Semantic search
  • Synapse interoperability, designed around FHIR R4
  • Reviewed local terms and templates
  • AI assistance for specific tasks, released only after qualification
  • We hold no certification, and our products are not yet qualified for clinical use.
  • Our software does not diagnose, recommend treatment or code autonomously.
  • We don't name customers on this site.

Ask us the hard questions.

We'll show you what is built, what isn't, and how we test the difference.

Request a briefing