THCOS
AI-Powered Multi-Agent Meeting Room — autonomous AI agents join your team meetings as intelligent participants. They take notes, assign tasks, track decisions, and follow up automatically. The smartest way to run productive meetings.
Platform Screenshots
Meeting Notes & Task Assignment
Auto-generated from AI participants
Overview
THCOS (The Company Operating System) is an AI-powered meeting platform that transforms how teams conduct meetings. Built entirely in-house by Lumora Build on Cloudflare Workers, THCOS introduces a revolutionary approach to team collaboration: autonomous AI agents join your meetings as intelligent participants, actively listening, taking notes, assigning tasks, and tracking decisions in real-time.
Unlike traditional meeting tools that simply record audio or provide basic transcription, THCOS creates a multi-agent environment where each AI participant has a specialized role. One agent might serve as the primary note-taker, capturing every important point and organizing them into structured summaries. Another agent acts as a decision tracker, identifying key decisions made during discussions and linking them to relevant context. A third agent serves as a task manager, extracting action items from conversations and automatically assigning them to team members with deadlines.
The platform addresses one of the most universal problems in workplace collaboration: the gap between what's discussed in meetings and what actually gets done. Studies show that 70% of action items from meetings are never completed, and nearly half of all decisions made in meetings are forgotten within a week. THCOS solves this by ensuring every meeting produces clear, trackable outcomes that persist beyond the meeting itself.
The Multi-Agent Meeting Concept
THCOS introduces the concept of "multi-agent meetings" — a fundamentally new approach to team collaboration. Instead of a human having to manually take notes, track decisions, and follow up on action items, multiple AI agents handle these responsibilities simultaneously and autonomously. Each agent is specialized for a specific function, creating a comprehensive meeting support system that operates in real-time.
The note-taker agent uses advanced natural language processing to identify key points, separate important decisions from casual discussion, and organize information into structured formats. It understands context, recognizes when topics change, and can differentiate between brainstorming ideas and final decisions. The resulting notes are comprehensive yet concise, capturing everything that matters without overwhelming readers with verbatim transcripts.
The task extraction agent listens for commitment language — phrases like "I'll handle that," "We need to follow up on," or "Can you look into" — and converts these conversational commitments into structured tasks with assignees, deadlines, and priority levels. This eliminates the common problem where action items are discussed but never formally tracked.
The decision tracker agent maintains a running log of all decisions made during the meeting, including the reasoning behind each decision, who supported or opposed it, and any conditions or dependencies associated with it. This creates an auditable record that can be referenced in future meetings when discussing the outcomes of previous decisions.
Built for Modern Teams
THCOS is designed for teams of all sizes, from small startups to large enterprises. The platform works with any meeting format — daily standups, weekly planning sessions, client presentations, board meetings, or brainstorming workshops. It integrates seamlessly with existing video conferencing tools and can be configured to join meetings automatically based on calendar events.
The platform supports multi-language meetings, with AI agents capable of understanding and processing conversations in multiple languages simultaneously. This is particularly valuable for global teams where participants may switch between languages or where non-native speakers are present. The AI agents provide real-time translation and can generate meeting summaries in the preferred language of each participant.
Security and privacy are built into every layer of the platform. Meeting recordings and transcripts are encrypted at rest and in transit, with configurable retention policies that comply with enterprise data governance requirements. Administrators can control which meetings are recorded, who has access to meeting summaries, and how long meeting data is retained.
Technology Stack
THCOS is built on Cloudflare Workers, providing global edge computing with sub-millisecond latency. This ensures that AI processing happens close to users, delivering real-time responses without the lag associated with centralized cloud processing. The platform uses Hono.js for the API layer, TypeScript for type safety, and Tailwind CSS for the frontend interface. The database layer uses Cloudflare D1 for persistent storage of meeting data, notes, and action items.
The AI capabilities are powered by Workers AI, Cloudflare's serverless AI inference platform, which allows THCOS to run sophisticated natural language processing models directly at the edge without requiring external API calls. This architecture provides both performance and privacy benefits, as meeting data never leaves the Cloudflare network.
The WebRTC infrastructure handles real-time audio and video streaming, while the multi-agent coordination system uses Durable Objects for stateful, consistent agent behavior across distributed edge locations. This ensures that all AI agents have a shared understanding of the meeting context, even when they're processing different aspects of the conversation.
Quick Facts
Key Features
AI Meeting Participants
Autonomous AI agents join your meetings as active participants — listening, understanding context, and contributing insights in real-time. Each agent has a specialized role in the meeting process.
Auto Meeting Notes
Automatic transcription and intelligent note-taking. Every discussion is captured, organized into structured summaries, and made searchable after the meeting ends. No more manual note-taking.
Task Assignment & Tracking
AI extracts action items from discussions and assigns them automatically with deadlines and follow-up reminders. Every commitment becomes a tracked task that won't be forgotten.
Multi-Agent Collaboration
Multiple AI agents with different specializations work together in every meeting — note-taker, timekeeper, decision-tracker — creating comprehensive meeting support.
Decision Tracking
Every decision made in meetings is recorded, timestamped, and linked to relevant discussions. Build an auditable history of team decisions for future reference.
Multi-Language Support
AI agents understand and process meetings in multiple languages simultaneously. Global teams can collaborate seamlessly with real-time translation support.
Understanding AI-Powered Multi-Agent Meetings
Why Traditional Meetings Fail
Meetings are fundamental to how teams collaborate, yet they remain one of the least efficient parts of the workday. Research from Harvard Business Review found that senior managers spend an average of 23 hours per week in meetings, with over half of that time considered wasted. The problems are well-documented: meetings run over time, participants lose focus, decisions aren't documented, and action items fall through the cracks.
The root cause of most meeting failures isn't poor scheduling or lack of preparation — it's the fundamental limitation of human attention. When you're in a meeting, you're either participating in the discussion, taking notes, tracking decisions, or following action items. You can't effectively do all four simultaneously. This cognitive bottleneck means that important information is lost in every meeting, regardless of how well-intentioned the participants are.
THCOS addresses this limitation by introducing AI agents that handle the information management aspects of meetings, freeing human participants to focus entirely on the discussion. The AI note-taker doesn't get distracted by side conversations or lose focus when topics shift rapidly. The task tracker doesn't forget to write down an action item because it was mentioned casually. The decision logger captures every commitment with perfect accuracy, creating a reliable record that humans can trust.
The Multi-Agent Architecture
THCOS uses a multi-agent architecture where each AI participant has a specific, well-defined role. This is different from having a single AI assistant try to handle everything. Specialized agents are more effective because they can be optimized for their specific function, and the system can scale by adding more specialized agents as needed.
- Note-Taker Agent: Responsible for real-time transcription and intelligent summarization. Identifies key points, organizes information by topic, and generates structured meeting summaries.
- Task Manager Agent: Listens for commitment language and extracts action items. Assigns tasks to participants, sets deadlines based on context, and sends follow-up reminders.
- Decision Tracker Agent: Monitors the conversation for decisions, records the reasoning behind each decision, and maintains an auditable decision log.
- Timekeeper Agent: Tracks meeting duration, monitors agenda items, and provides time alerts to keep discussions on track.
- Sentiment Analyst Agent: Analyzes participant engagement and mood, providing insights about team dynamics and meeting effectiveness.
Real-Time Processing at the Edge
One of THCOS's key technical advantages is that all AI processing happens at Cloudflare's edge network, providing sub-millisecond latency for real-time meeting support. This means that notes appear on screen as people speak, tasks are assigned instantly when commitments are made, and decisions are logged the moment they're agreed upon. There's no lag, no delay, and no post-processing required.
This edge-first architecture also provides significant privacy benefits. Meeting audio and transcripts never leave the Cloudflare network, which is particularly important for organizations handling sensitive information. The data is processed in isolated worker instances that are destroyed after each meeting, ensuring that no residual data persists beyond the meeting itself.
How THCOS Transforms Meeting Outcomes
The most immediate impact of THCOS is the elimination of the "meeting amnesia" problem. Teams using the platform report that action item completion rates increase dramatically because tasks are captured automatically and assigned with clear ownership. Decisions that would previously have been forgotten or misremembered are now documented with full context, making it easy to reference past decisions when discussing current issues.
Beyond the immediate benefits of better documentation, THCOS creates a knowledge base of meeting history that compounds in value over time. New team members can review past meetings to understand how decisions were made, why certain directions were chosen, and what commitments were made. Project retrospectives become data-driven rather than relying on imperfect human memory.
The platform also provides meeting analytics that help teams improve their meeting culture over time. Managers can see which meetings are most productive, which participants contribute most effectively, and where time is being wasted. This data-driven approach to meeting improvement helps organizations gradually build more efficient meeting practices.
Integration with Existing Workflows
THCOS is designed to integrate seamlessly with existing meeting workflows. It can join meetings automatically based on calendar events, or participants can manually invite AI agents to ad-hoc discussions. The platform integrates with popular project management tools, automatically creating tasks in systems like Jira, Asana, or Trello when action items are extracted from meetings.
Meeting summaries and action items are distributed automatically after each meeting, with different participants receiving customized summaries based on their role and the items relevant to them. This ensures that everyone gets the information they need without being overwhelmed by details that don't concern them.
The platform also supports asynchronous meeting participation, where team members who couldn't attend a meeting can review the AI-generated summary, see the decisions that were made, and understand the action items that were assigned. This is particularly valuable for distributed teams working across time zones.
Security and Compliance
THCOS is built with enterprise-grade security from the ground up. All data is encrypted at rest and in transit, with configurable retention policies that comply with GDPR, CCPA, and other data protection regulations. Administrators have full control over which meetings are recorded, who has access to meeting data, and how long information is retained.
The platform supports single sign-on (SSO) integration, role-based access controls, and audit logging for all data access. Meeting recordings can be configured to be automatically deleted after a specified period, and participants can opt out of recording for sensitive discussions.
Use Cases
Daily Standups
AI agents track what each team member accomplished, what they're working on, and any blockers. The resulting summary provides a clear record of daily progress that can be referenced in sprint reviews.
Client Presentations
During client meetings, AI agents capture requirements, decisions, and follow-up items. The resulting summary can be shared with the client as a professional meeting recap, demonstrating attention to detail.
Brainstorming Sessions
AI agents organize brainstorming output by theme, identify the most popular ideas based on discussion frequency, and create a structured list of next steps for idea development.
Board Meetings
Board meetings require precise documentation of decisions and commitments. THCOS provides formal, auditable records that meet governance requirements while reducing the administrative burden on board secretaries.
Global Team Meetings
Multi-language support ensures that all participants can follow the discussion regardless of their native language. Meeting summaries can be generated in each participant's preferred language.
Compliance Meetings
Regulatory and compliance meetings require detailed documentation. THCOS provides comprehensive records that can be used for audit purposes, with timestamps and participant attribution for every decision.