OsteoAI is a clinical reasoning assistant. Paste a presentation — age, complaint, findings, history — and pick how you want it reasoned about. It works through differentials, treatment plans and case simulations with you, citing what it leans on¹.
OsteoAI doesn't summarise. It walks the case the way you would — narrowing differentials against the presentation, flagging what would change its mind, and writing every step to a graph you can revisit.
Age, complaint, mechanism, findings, history. Plain prose — no fields, no required template. Attach imaging or notes if you have them.
Five modes, five intents. The mode changes the prompt, the citation depth and what the assistant withholds. Switch mid-case.
Every concept the model leans on is added as a node; every chain becomes an edge. New concepts emerge from related ones — so you see what was derived from what.
Each mode is a different reasoning intent — picked above the composer or via the sidebar. You can switch mid-conversation; the graph carries over.
The assistant walks the case end-to-end — interpretation, working hypothesis, what would change its mind. Conversational, citation-grounded.
Best for: complex or atypical presentationsA ranked, weighted list of differentials with the evidence for and against each — plus the single test or sign that would discriminate the top two.
Best for: focused diagnostic workA virtual patient with internal state. Take a history, ask for tests, propose treatment — the simulation responds with realistic, sometimes ambiguous findings.
Best for: training, exam prep, studentsA 4–8 session plan with technique selection, dosage rationale, between-visit homework and explicit re-assessment markers. Editable; you own the final plan.
Best for: post-diagnosis planningA plain conversational mode. Look up a technique, ask about anatomy, discuss a paper, sense-check a colleague's reasoning. No case structure imposed.
Best for: research, study, side questionsLong-running notes per patient. The assistant carries forward findings, treatment history and your reasoning across sessions — so the third visit isn't starting from blank.
In private beta — request accessConcepts the assistant leans on become nodes. The chain of reasoning becomes edges. New concepts emerge from the related ones rather than arriving from outside — so you can see which prior cases informed today's working hypothesis.
Drivers are dense and filled. Symptoms are bare. Amplifiers come in amber. Uncertainty is rendered as a dashed outline — the graph admits when it isn't sure.
Driver — filled sage, dense bloom
Amplifier — amber, medium bloom
Compensation — outlined
Symptom — dot inside, no dendrites
Uncertain — dashed outline
Clinical software earns trust by what it refuses to do, not what it claims. This page is more useful before you've signed up than after.
We considered three tiers with checkmark grids. We threw it out. You're either a practising clinician or a student — both deserve the same product.
500 reasoning queries / month. All five modes. Knowledge graph. Citation export. Locked in for the duration of beta — no surprise re-pricing.
If yours isn't here, write to us. We answer ourselves.
A large language model with domain-specific reasoning architecture and a curated osteopathic knowledge base. The model is guided by clinical reasoning rules built with practising osteopaths — not general medical prompting.
Don't paste it. The product is built for de-identified presentations. We don't ingest, store, or train on patient records, and we strip obvious identifiers on input as a backstop. If you need a PHI-compliant deployment, talk to us about a private instance.
Yes — and we built Case Simulation mode largely for you. Student pricing is €9/month with a verified institutional email. The simulations are designed to be punishingly realistic, including ambiguous findings and the occasional red herring.
Beta pricing is locked in for as long as your subscription is continuous. We expect to add tiers — solo, clinic, institution — but beta clinicians stay on the price they signed up at.
Domain. Voice. Mode-specific prompting. A knowledge graph that learns your caseload. And a deliberate refusal to do things outside scope — it won't write patient-facing copy, won't diagnose, won't recommend specific medication.
OsteoAI is built in active dialogue with clinicians — not handed down from engineers to users. The product gets sharper every time a case reveals a gap in the reasoning.
Beta members shape what gets built next. If your caseload reveals a reasoning gap, that gap becomes a prompt fix — usually within days.
Beta pricing holds for the life of your subscription. No re-pricing when we leave beta. You backed it early — that counts.
OsteoAI grows the way good clinical tools always have — one colleague recommending it to another. No ad spend. No growth hacking.
50 free queries, no card required. All five modes from the first session. If the reasoning doesn't hold up against your clinical judgment — you lose nothing. If it does, you'll want it in every consultation.
50 free queries. No card. All five modes unlocked from the first session. Open the assistant and try a case you already know the shape of — see if its reasoning lines up.