AAI Labs · Healthcare
Three products. One promise: less admin, more care.
Lithuanian healthcare runs on paperwork. Doctors document instead of listening. Reception staff keep dozens of scenarios in their heads. Staff search five systems to answer one patient question.
We build AI that fixes the operational side of healthcare — on your servers, in your language, with every claim backed by a citation.
Working with
The problems
A decision involves leadership, clinical leads and IT. Each of them sees a different problem. Pick yours.
Family doctor
The problem today
Listening, examining, deciding and writing all compete inside a short visit. The patient's history sits in separate records. Unfinished E025 records move to the evening.
What changes
A structured E025 draft and suggested codes to review and sign, and the patient's history summarised before they walk in.
Reception operator
The problem today
Has to follow 20+ clinical scenarios, and the questions are easy to forget mid-call.
What changes
The scenario's questions appear on screen, and any required one still missing is flagged before the call ends.
Reception manager
The problem today
Can't see what happens on calls. Training is informal and turnover is constant.
What changes
Every call has a record: scenario, answers, duration, outcome. Training becomes evidence-based.
Legal, quality and admin staff
The problem today
Internal procedures, descriptions and the law are scattered. It's hard to find which procedure applies or who is responsible, and legal changes are easy to miss.
What changes
One chat over the institution's documents and the law, every answer with its source, and agents that warn when upcoming legislation affects internal procedures.
Director or head doctor
The problem today
It's hard to see where staff time goes, or to prove the quality of care when a patient or an inspector asks.
What changes
Clear indicators of time saved and procedures followed, and a record of every call and document.
IT and CISO
The problem today
Patient conversations and records can't be sent to a third-party cloud.
What changes
Deployed on the institution's own servers, with named-user access, role permissions and an audit log.
Patient
The problem today
Long phone waits, and the quality of service depends on who answers and which language the patient speaks.
What changes
The same protocol for everyone, a faster call, and a lower chance of being sent to the wrong place.
The products
Each one supports a professional who stays accountable. Each one runs on the institution's servers.
Doctor Assistant
Listens to the consultation and drafts the E025, summarises the patient's history before the visit, and suggests the codes. The doctor reviews and signs.
Reception Assistant
Helps reception staff follow the polyclinic's scenarios during a call. It shows the questions, ticks them off, and reminds the operator of anything still missing.
Merkys.ai
Staff ask in plain language and get an answer from the institution's own documents and the law, with the source. Agents follow upcoming legislation and check that internal procedures keep up.
How we work
All three products run inside the institution. No patient data leaves the building, and no third-party cloud is involved.
How we deploy →Hand-checked evaluations, including where the model falls short: 150 of 150 answers grounded in source text, 25 of 25 unanswerable questions refused.
See the numbers →Doctor Assistant is part of a European research project; Merkys builds on our own research and benchmarks.
Where it comes from →Tell us which product interests you, and we'll get back to you.
