Radiology dictation · end-to-end encrypted
RadDictate turns your speech into clean findings and conclusions inside any web-based RIS or HIS — quietly proofreads the report for the slips fatigue makes, and stores your history so encrypted that even we can't read it.
› Developed by a practicing radiologist › No install on hospital servers › Works on hospital networks
Workflow
No copy-paste, no separate window to babysit. RadDictate lives inside the system you already report in.
A floating pill follows you around your RIS in Chrome. Or dictate on your phone while the report fills in on the PC — the phone is just a better microphone.
The AI turns natural dictation into clean report language, following your templates and house style — not a generic transcript.
Findings go to the findings field, the conclusion to the conclusion field, inserted exactly where your cursor was — mid-sentence if that's where you clicked.
Second read
Dictation slips are quiet: right becomes left, the sigmoid becomes the transverse, and nobody notices until it matters. RadDictate reads the finished report against itself and flags internal contradictions — it never edits silently.
Laterality, anatomic sites, contradicting statements between findings and conclusion — the unintended mistakes of a long reporting list.
A flagged report is still your report. The AI marks the discrepancy and steps back; the radiologist who signs it stays in charge.
Organizations can add their own checks, templates and reporting style — the second read learns your department's rules, not someone else's.
Zero knowledge
Most medical software promises not to look at your data. RadDictate is built so that looking is impossible: reports are encrypted on your device with a key derived from your password — a key we never see.
Your password is stretched into an encryption keyPBKDF2 · 600,000 rounds
Each report is sealed before it leaves the browserAES-256-GCM
Ciphertext. That's all.
Everything — including your audio — travels over TLS encryption, end to end, on every client.
Speech-to-text must momentarily process audio. It runs on Google Vertex AI with zero data retention configured — audio is never stored, never used for training.
History auto-deletes after 14 days by default — set 1–60 days or keep it forever. Old reports don't pile up waiting to leak.
Everything else
Your report structures, your phrasing. The AI formats to your templates — normal studies read the way you'd have typed them.
Reporting on a workstation with no mic — or one you don't trust? Dictate into your phone; the finished report appears in the browser on the PC, encrypted the whole way.
Reports sync across your extension, phone and desktop — end-to-end encrypted — so the study you reported on the go is at your workstation before you are.
For departments that cannot send audio out at all, our on-premise desktop engine transcribes locally. No internet. No cloud. Nothing leaves the room.
Hospital proxies, shared IPs, thin bandwidth — RadDictate is engineered for the networks radiology actually runs on, not a Silicon Valley office.
Per-device sign-in limits and usage quotas keep group licences honest — a feature your administrator will appreciate.
For organizations
Why this exists
"The dictation systems I admired come at prices radiologists in my part of the world simply cannot afford. I refused to accept that good reporting tools are a rich-hospital privilege — so I built one that matches them, exceeds them where it matters, and costs a fraction as much."
— The founding radiologist · OUI Technologies, Pakistan
Questions
14 days · full features · no card