Speech-to-Action
Control your workflow with your voice.
Speech-to-Action is not dictation. It triggers clinical automations via voice while your eyes stay on the images. Agentic radiology, only possible on a unified cloud-native platform.
How Sirona is Different
Speech is the natural clinical interface.
In radiology, the reading position is biomechanically constrained — your eyes must stay on the images, and your hands are occupied with input devices or measurement tools. Dictation solved the output problem: you can speak your report instead of typing. Speech-to-Action extends the voice interface to the entire workflow. You can say 'load the prior chest from 2023' without interrupting your read. You can say 'run the AI algorithm on this region' and an agent acts. This is about clinical efficiency: focus on the images, reduce context switching, spend more time on clinical judgment.
What speech-to-action enables
Voice-driven control across every layer of the reading workflow — from viewer navigation to AI agent invocation to worklist management — powered by a unified data model and real-time event monitoring.
Real-Time Voice Commands
Control the viewer, load comparisons, navigate series, adjust windows, and manage the worklist entirely through voice — with real-time response because everything runs on a unified platform.
Explore Reporter
RadGPT Voice Agents
Invoke agentic AI through natural language. 'Run the lung algorithm on this region,' 'show prior findings in this area,' 'flag discrepancies with the prior report' — agents understand context and act intelligently.
Meet RadGPT
Speech Event Monitoring
Voice commands flow through the same unified event stream as viewer clicks and reporter edits — making it possible for agents to understand and respond to what the radiologist is doing right now.
Event stream
Unified Data Model
Every voice command has access to the full reading context — what's visible in the viewer, what's been dictated, what the prior reports say, what clinical alerts exist — because it's all in one place.
Cloud-native
Cloud-Native Responsiveness
Voice commands execute in real time across a cloud-native platform. No latency from cross-system API calls, no waiting for vendor coordination — your command is your platform acting.
Cloud-native
Practice-Configurable Actions
Define custom voice commands that trigger your workflows — 'route to attending,' 'mark for conference,' 'create a prior summary' — coming as the Amplify SDK rolls out.
Amplify SDKThe sound of agentic radiology
Interface
Your voice is the fastest input device
Keyboard and mouse were designed for general-purpose computers. The reading position was never a general-purpose computer — it's a constrained, high-attention environment where your eyes belong on the images. Voice removes the friction. Every command the workflow supports can be spoken, with real-time response.
Hands-free viewer navigation
Worklist triage by voice
Measurement and comparison commands
Stays in focus — eyes never leave the images
Context
The agent knows what you mean
Voice commands on Sirona aren't keyword matches. The agent knows which study is open, which series you're on, what you've already dictated, and what priors exist. 'Pull the prior' becomes an action grounded in the current session, not a dialog box asking you which prior.
Session-aware intent understanding
Access to viewer, reporter, worklist, and AI state
Disambiguates by what's currently on screen
Grounded in the unified data model
Automation
From dictation to agentic voice
Every voice command can trigger an agentic workflow — not just a single action. 'Summarize priors' doesn't only open a prior; it reads the prior report, extracts the relevant findings, and presents them inline. 'Flag for conference' doesn't just tag the study; it generates a review packet with key images.
One voice command triggers a chain of actions
RadGPT agents execute in seconds, not minutes
Practice-configurable command library
Logged on the unified event stream
Architecture
Only possible on a unified platform
Legacy stacks can't do real speech-to-action because a voice command crossing the viewer, reporter, worklist, and AI vendor is a coordination problem. Sirona owns the full stack, so the command is local to a single platform. That's the architectural prerequisite.
Unified data model + event stream
No cross-vendor API coordination
Cloud-native compute keeps latency low
Only cloud-native, unified platforms qualify
Hands-free
clinical workflow control via voice
Real-time
command execution as you read
1
unified platform required — Sirona
How voice changes the read
Why Agentic AI Requires RadOS
Watch as Dr. Mark Longo demonstrates the power of a new paradigm in radiology AI. Welcome to the age of agentic, embedded, multimodal, real-time, clinically aware AI assistants.
Dr. Mark Longo
Chief Technology Innovation Officer, Sirona
Read the partnership announcement “Through Sirona's platform, Everlight will be able to build and deploy AI-powered automations across clinical, administrative, and operational workflows. This partnership will fundamentally transform how our radiologists practice, and position Everlight at the forefront of AI-enabled diagnostic medicine globally.”
Jeff Oakman
Global Chief Operating Officer, Everlight Radiology
FAQs
Isn't this just dictation?
Which commands are supported?
What microphone do I need?
Is this FDA-cleared?
How is this different from Dragon Medical?
Does this require RadGPT?