Platform
We start from the decision, not the data.
Healthcare doesn’t lack information. It has an institutional knowledge gap: what an organisation knows, but can’t find, trust or prove at the moment someone must decide. That gap is what we build the platform around.
Hospitals, insurers, drug makers, researchers: each gets a custom platform, designed and built together with the organisation that runs it.
Why now
The institutional knowledge gap is widening.
- 0.0 yrs
until half of clinical guidelines studied were outdated
Shekelle et al., JAMA, 2001 - 0%
of physicians report care delays from prior authorisation
AMA physician survey, 2025 - 0%
of trial protocols now need at least one amendment
Tufts CSDD, 2024 - 50–0%
of AI answers to medical questions not fully supported by their own sources
Nature Communications, 2025
How we think
Six things we won’t compromise on.
- Decision firstStart from the moment someone must act. Work back to the institutional knowledge it needs.
- Experts decideAgents propose. Your named experts certify. Nothing runs unsigned.
- Show the basisSource, version, approver, scope. On every answer.
- No is an answerNothing certified applies? It says so, and names who decides.
- Change is the workWhen a source changes, everything built on it is flagged.
- Yours aloneYour own certified layer. Never pooled, never shared.
How that’s different
Most approaches store information. We certify decisions.
- What’s storedDocuments and data
- Who decides what’s trueThe model, or whoever wrote it last
- When a source changesFound out later, if at all
- When it isn’t sureIt guesses, fluently
- Looking aheadReports on what already happened
- ProofA transcript
- Across organisationsOne shared model or pool
- What’s storedDecision-ready knowledge, each item with its basis
- Who decides what’s trueYour named experts
- When a source changesEverything built on it is flagged
- When it isn’t sureIt says no and names who decides
- Looking aheadPredicts, simulates and acts, within what your experts signed
- ProofA passport on every answer and every refusal
- Across organisationsYour own layer, under your governance
Kinds of platform
One context. 24 kinds of platform.
What is true, according to whom, as of when?
From describing to acting
Every kind reaches as far as your experts sign: from what happened, to doing it for you within the limits they set.
- DescribeWhat happened, by your certified definitions.
- DiagnoseWhy, by the causes your experts accept.
- PredictWhat will happen, on models validated for you.
- PrescribeWhat to do, on the protocol in force.
- ActDone for you, only inside what your experts signed.
Where it applies
Six parts of healthcare. One layer under all of them.
Criteria change faster than reviewers can read them.
Every criterion versioned, every decision traced to the one in force, and policy changes simulated before they ship.
- Learning from outcomes
- Medical policy library
- Benefit rules
- Member explanations
- Rising-risk members
- Cost and demand forecasts
- Policy change impact
- Next best action
- Benefit design options
- Prior authorisation
- Utilisation review
- Appeals
- Care management programmes
- Payment integrity
- Regulatory timelines
- Member data sharing
- Payer and provider exchange
- Provider network rules
- and whatever comes next
Examples, not a limit. Healthcare adds new rules, evidence and products every year. If a decision rests on knowledge someone must stand behind, the platform can carry it.
See full coverage
| For | Know | Foresee | Act | Guard | Connect |
|---|---|---|---|---|---|
| Payer |
|
|
|
|
|
| Provider |
|
|
|
|
|
| Pharma |
|
|
|
|
|
| Life sciences |
|
|
|
|
|
| Personal health |
|
|
|
|
|
| Health tech |
|
|
|
|
|
In principle
The same engine under every decision.
- CaptureReads your sources: guidance, labels, policies, protocols, contracts, the knowledge in your experts’ heads.
- ProposeAgents draft each item of context, with where it came from.
- CertifyYour named experts approve, edit or reject. Nothing runs unsigned.
- Keep currentWhen a source changes, everything built on it is flagged for re-certification.
- DeliverTo people, workflows and other AI, at the moment of the decision.
- LearnOutcomes flow back. Your experts certify what should change, and the rules improve.
- Prove and refuseEvery answer carries its passport; with no certified basis, the answer is no, on the record.
What every answer carries.
- The claim, or the action
- Its source, and the version
- Who approved it, by name
- Where it applies, and where it doesn’t
- When it must be reviewed
A first engagement
Small first. Then wider.
- 01Scoped togetherWe agree the problem, the length and what success looks like before anything starts.
- 02Your data lastYou see it working on made-up data before any of yours is involved.
- 03Your experts decideNothing goes live without the approval of your named experts.
- 04Yours to leaveIf you leave, your certified context is archived, then deleted after a set retention period.
You bring one problem, a named lead and the experts who sign off. The research behind the problem: the Knowledge Cost Index.
Start with one decision.
We shape the platform around it with your experts, on your own certified context, and grow from there.