Orchestrating and fine-tuning LLMs

🎙️ He Taught AI How to Have Manners — Meet David Petrou of Continua AI

What if your next group chat had an extra participant — one that listens, understands the social context, remembers what you said last week, and even knows when to stay quiet? In today’s episode, host Dietmar Fischer sits down with David Petrou, founder and CEO of Continua AI, to explore the emerging world of Social AI — intelligent agents designed not just to talk, but to collaborate inside group chats.

David, formerly at Google and part of the original Google Glasses team, has spent decades thinking about how humans and machines interact.

With Continua, he’s building the world’s first truly human-aware AI that can join your Discord, iMessage, or Google Message conversations and behave like a socially intelligent teammate. This isn’t a chatbot — it’s an AI that understands when to talk, when to listen, and when to help.

📧💌📧

Get my Newsletter

Tune in to get my thoughts and all episodes, don’t forget to subscribe to our Newsletter: 👉 https://beginnersguide.nl
📧💌📧

Get ready for a deep dive into social intelligence, etiquette in AI systems, agentic actions, and the future of communication where AI participates naturally alongside humans.

💡 What You’ll Learn in This Episode

• Why Social AI is the next big evolution beyond traditional chatbots

• How Continua trains AI to understand timing, tone, context, and social cues

• Why David believes text messaging with AI will reach a billion users

• The engineering challenge behind teaching AI “manners” and “machine etiquette”

• How AI group chat agents improve communication, planning, and collaboration

• The real use cases: debugging code, planning trips, updating documents, running games, and summarizing information

• How Continua’s multi-model architecture orchestrates LLMs, fine-tunes, and intent classifiers

• Why Social AI is surprisingly safe — and why today’s fears don’t match the technical reality

• The leadership perspective: how to integrate AI thoughtfully without overwhelming teams

• Where Social AI is heading next: meetings, real-time participation, contextual computing, and agentic actions like shopping

This episode is packed with insights for anyone interested in AI agents, human–AI collaboration, team communication, or the future of intelligent digital assistants.

📌 Quotes from the Episode

• “We had to break the LLM’s brain and teach it social etiquette: when to talk, when to listen, and when to stay quiet.”
• “Traditional chatbots operate in single-player mode — Continua is built for multiplayer conversation.”
• “There are problems beyond our ability to solve directly — the real ingenuity is creating something that can learn how to solve them.”

• “Introducing a foreign intelligence into human group dynamics is one of the most fascinating problems in AI.”

• “Text messaging with AI will be the next form factor to hit a billion users.”

• “Language itself is the interface. You don’t need menus. You just tell the AI how you want it to behave.”

⏱️ Chapters

00:00 David Petrou’s Origin Story & Early Fascination with AI
04:51 Why Social AI Matters: From APIs to Human-Aware Group Agents
09:12 Teaching AI Social Etiquette: When to Talk, Listen, or Stay Quiet
16:11 Inside Continuum: Multi-Model Architecture, Fine-Tuning & Real Use Cases
24:05 Social AI in the Real World: Planning Trips, Debugging, Collaboration & Automation
35:01 The Future of Social AI: Meetings, Agentic Actions, Leadership & Ethical Considerations

🧑‍💼 About Dietmar Fischer

Dietmar is a podcaster and AI marketer from Berlin. If you want to know how to get your AI or your digital marketing going, just contact him at argoberlin.com

🔗 Where to Find the Guest: David Petrou

• Website: continua.ai
• LinkedIn: David Petrou
• Instagram: David Petrou

🎵 Closing Credits

Music credit: “Modern Situations” by Unicorn Heads