The AI copilot that sits in on every eye exam
EyeAI listens to the consultation in real time — or to an uploaded transcript or test results — and quietly surfaces differentials, red flags, the right exam steps and ICD codes. Every suggestion is cited to EyeWiki, so you can verify in one click.
Connected to your LLM provider.
Live transcript
Patient
Yesterday my right eye suddenly went dark, painless. I'm 68.
Doctor
Any headache, jaw pain when chewing, or scalp tenderness?
Patient
Some headache this week and my jaw aches when I eat.
Copilot
Possible giant cell arteritis with CRAO — start high-dose steroids on suspicion; urgent ESR/CRP.
EyeWiki: Giant Cell ArteritisWorkflow
Three inputs, one assistant
Capture the visit however suits your clinic. The copilot does the rest.
Capture the encounter
Record the live conversation, upload an audio file, or paste an existing transcript or test results. Cloud speech-to-text handles the listening.
Get grounded support
The copilot reasons over the dialogue, retrieves the relevant EyeWiki articles, and surfaces differentials, red flags, questions and exam steps.
Document in one click
Generate a structured SOAP note and suggested ICD-10 codes from the same transcript — review, edit, and drop into your record.
Capabilities
Built for the rhythm of a busy clinic
Decision support that stays out of the way until you need it.
Live differentials
A ranked, evolving differential that updates as the history unfolds — each with a one-line rationale.
Red-flag alerts
Vision- and life-threatening presentations (CRAO, angle-closure, GCA, detachment) are surfaced first, with urgency levels.
Exam checklists
The next examination steps tailored to the presentation, so nothing high-yield is missed.
EyeWiki citations
Every clinical claim links back to its EyeWiki source for instant verification.
SOAP + coding
One-click SOAP notes and suggested ICD-10 codes drafted from the transcript.
Clinician-in-command
Assistive by design. The copilot proposes; you decide. Nothing is auto-filed.
Knowledge layer
Beyond plain RAG: a hybrid, cited knowledge core
Following 2026 best practice, the copilot uses a hybrid tiered design. A compact golden core of ophthalmology is preloaded into the model (cache-augmented generation) for instant, retrieval-free reasoning — while an agentic retrieval tool pulls exact, citable EyeWiki passages on demand.
- Zero-latency reasoning over common presentations
- On-demand retrieval for precise, source-linked detail
- Citations the model actually retrieved — never fabricated
- Extend the corpus anytime with the EyeWiki ingest script
+ 1598 more articles, indexed for retrieval
Decision support — not a diagnosis
EyeAI is an assistive tool. It does not diagnose, prescribe, or replace clinical judgment. Every suggestion is a prompt for the clinician, cited so it can be independently verified. The treating clinician remains fully responsible for all decisions and documentation.
Bring the copilot into your next exam
No setup, no patient data stored. Try a live demo with sample cases in under a minute.
Open the copilot