Continua Raises $8M to Put LLMs in Group Chats

In early 2023, David Petrou — a founding engineer on Google Goggles and Google Glass — left Google to build Continua, a startup that brings large language models into group chats on SMS, iMessage and Discord. This week Continua announced an $8 million seed round led by GV with participation from Bessemer and angels.

Funding and founder momentum

Investors backed Petrou’s vision early. GV’s Erik Nordlander said they invested before the idea was fully formed, betting on Petrou’s engineering track record and the market opportunity for social-aware AI in everyday group coordination.

What Continua actually does

Continua’s agents join group conversations and step in only when useful. They summarize plans, set reminders, create polls, add calendar invites, and generate shared documents with checklists and to-dos. If someone forgets a detail, they can privately message the agent for a quick lookup.

• Automated reminders, polls, and calendar invites
• Shared documents and checklists generated from chat context
• Private direct messages for forgotten details or sensitive lookups

At first glance this looks simple, but Petrou explains the core technical challenge: most LLMs are trained for one-on-one assistant dialogues. Continua needed to teach models social intelligence — when to speak, when to be silent, and how to follow multiple conversational threads without derailing the group.

Petrou describes the work as having to “break the LLM’s brain” and re-tune its instincts so the agent behaves naturally among humans. Users can also tell the agent to hang back or specifically invoke it when needed, keeping control in human hands.

Continua enters a growing field — Meta and startups like Hey Umai are experimenting with conversational agents — but Petrou positions his product as tuned specifically for multi-person interaction rather than solo assistant use-cases.

Why this matters

Group chat is how people coordinate everything from family trips to product launches. Removing repetitive tasks and surfacing decisions can save time and reduce friction. Continua’s approach could change how teams plan events, manage volunteer groups, or coordinate shift schedules — turning chaotic threads into structured action.

• Event planning: auto-agenda and RSVPs
• Corporate coordination: compiling meeting notes and action items
• Public sector and emergency response: rapid, contextual summaries across many participants

For organizations considering group-chat AI, the questions are practical: how well does the agent handle messy conversation? What are the privacy and audit controls? How do you measure reduced friction or saved time? Investors see monetization paths like paid features for planning or booking, but real value hinges on trust and fit.

QuarkyByte’s approach is to analyze these trade-offs with neutral evidence: benchmark social-intelligence, model privacy exposure, and design pilots that measure real-world productivity gains. That way, enterprises and public organizations can adopt group-chat agents where they add clear value — not just novelty.