Agent · 100% on-premise · our own reasoner

Diagnostic Chat

cited evidence from clinical cases and papers with cited evidence, on local server.

Request demo View the full agent network

What this agent does

Diagnostic Chat is the conversational clinical agent of Dental Brain. It answers the dentist’s questions with cited evidence from more than 18,000 indexed scientific papers and 113,000 validated clinical cases. All inference runs on the clinic’s local server, without sending data to the cloud.

Why it matters

A clinic adopting conversational AI for diagnostic support needs three things the cloud cannot deliver: low latency on critical queries, full sovereignty over sensitive clinical data, and traceability of the bibliographic source behind every answer. Diagnostic Chat solves all three on the same server the clinic already owns.

How it works

The dentist types (or dictates) the query. The agent retrieves relevant fragments from the clinical memory and the clinical cases using contextual search. Dental Brain’s proprietary LLM — trained specifically on odontological literature — synthesizes the response and returns it with verifiable citations. Average latency: 1.8 seconds.

Integration with the clinical workflow

Diagnostic Chat does not live in isolation. When the query involves a radiograph, it collaborates with Librarian Dental for deeper evidence retrieval. When it leads to a prescription, it passes the context to Vademecum and Pharmacy to verify interactions. Everything inside the same server, no cloud.

Autonomous decisions it makes

Retrieve relevant fragments from the clinical memory before answering
Cite verifiable bibliographic sources on every clinical answer
Delegate to Librarian Dental when the query needs deep paper search
Pass context to Vademecum and Pharmacy when the conversation involves prescription
Refuse to answer when available evidence is insufficient

Inputs and outputs

Receives

· Textual query from the clinical professional
· Voice dictation (processed locally)
· Active patient context (optional)
· Linked DICOM image (when applicable)

Produces

· Clinical answer with bibliographic citation
· Related papers list (Top-K)
· Suggested routing to other agents
· Confidence level of the response

Production metrics

1.8s

Average latency

18,000+

Indexed papers

113,000+

Clinical cases

Collaborates with

Tech stack

Model

Proprietary LLM trained on real dentistry data

Execution

Local on-premise GPU (no compromise between modules)

Latency

Immediate response

Privacy

Data never leaves the clinic server

Frequently asked questions

Does Diagnostic Chat replace the dentist's judgment?+

No. It is a support agent: it provides cited evidence to inform the decision, but the clinical decision is always the professional's. Every answer includes confidence level and backing papers.

Does my conversation data leave the clinic?+

No. Both the LLM and the paper database are deployed on the local server's GPU. There is no calls to external cloud services nor any external service. Native GDPR-compliant and healthcare-data sovereign.

What if evidence is contradictory across several papers?+

The agent presents relevant positions with their respective citations and leaves the assessment to the dentist. It does not fabricate a consensus that does not exist in the literature.

Can I dictate the query by voice?+

Yes. voice dictation running locallyalso runs locally, so your voice never leaves the server. Particularly useful during consultation when your hands are busy.