Webinar · 18 Nov 2026
Let's Dialog: From Model to Agent: Building a live agent
Speakers
Angus Addlesee
Senior Research Scientist
Ajeet Prabu
Agent Designer
Peter McKee
Developer Relations Manager
Register now
Join us live on November 18!
Loading form… This form needs JavaScript to show here. You can still fill it in on HubSpot. This form can't load here. You can still fill it in on HubSpot.
Open this form on HubSpotThanks! You'll receive an email confirmation with the webinar details. See you there!
This webinar has ended
In Part 1, the researchers behind Dialog-RSN-1 showed how it hears a call. In Part 2, we put it on one.
Peter McKee, Angus Addlesee and Ajeet Prabu will build a real-time dialog agent on Dialog-RSN-1 from scratch, live. We’ll connect to the API, write the prompt, add a tool, and take a call, showing the latency as we go. You’ll see how the agent handles a caller who pauses mid-sentence, cuts in, or calls from a noisy street.
Then we’ll try to break it. Bring your hardest caller behavior to the chat, and we’ll fix whatever breaks on camera. Expect a working session with real debugging, and plenty of room for your questions.
This is Part 2 of Let’s Dialog: Building the Next Voice Agents. Catch Part 1 live on Nov 4 to see how the model works before we build on it.
What you’ll take away
- Go from API key to a live agent. The steps to connect, prompt and put a real-time agent on Dialog-RSN-1, in the order you’d do them yourself.
- Prompt the turn-taking. How to tell the agent to wait out a pause or respond quickly, and hear the difference on the same call.
- Watch it act, not just talk. A tool call on screen, from the caller’s request to the result coming back.
- Learn what breaks and how to fix it. Interruptions, background noise and prompt edge cases, debugged live.