The Source of Truth for Every Meeting

Context

[START OF SECTION] Brainforge’s async team relied on Zoom for decisions, but those conversations quickly got buried. Transcripts were scattered across Slack, Notion, and inboxes—and Zoom’s built-in recordings were rarely touched or reused. Valuable knowledge was getting lost, and repeat work crept in.

The team needed a centralized, AI-searchable source of truth that could turn every Zoom meeting into accessible knowledge without adding new tools or licenses. The goal was to auto-ingest recordings, summarize key moments, and let teammates ask questions like “What did we decide on pricing last month?” The platform would become the go-to knowledge hub for leadership, product, and ops teams. [END OF SECTION]

Challenge

[START OF SECTION] Leadership kept getting pinged for repeat context. Finding past decisions meant chasing Slack threads or rewatching full meetings—a time-consuming and frustrating process. Previous attempts to fix this with Slack-based AI agents failed due to low adoption.

Key pain points included:

• Endless cycle of Slack ping-pong, repeated questions, and buried meeting notes
• No reliable way to turn raw Zoom recordings into trusted, searchable knowledge
• Leadership context was inaccessible without slowing them down
• External tools required new licenses and created adoption friction

What the team needed was simple: a private, reliable way to find past decisions fast. [END OF SECTION]

Solution

[START OF SECTION] Brainforge built a streamlined, private interface that auto-ingests Zoom recordings and unlocks instant meeting recall. The solution combined several technologies into a cohesive platform:

Data Ingestion Pipeline: Auto-recorded Zoom calls are pulled into Amazon S3 and summarized via Python workflows in Windmill. This creates an automated pipeline that requires no manual intervention.

Knowledge Storage: Transcripts and metadata are stored in Supabase, making them searchable via natural language queries. The database structure supports fast retrieval and context-aware search.

User Interface: A custom chat interface mimics the familiar Zoom experience but adds context-aware search across Slack and meeting history. Users can ask questions and get instant answers from past meetings.

Screenshot showing the platform interface with two panels: a meeting details view on the left (showing participants, recording info, and summary/transcript tabs) and an AI Meeting Assistant chat interface on the right where users can ask questions about meeting content and get summarized decisions.

Tools Used:

• OpenAI for natural language processing and summarization
• n8n for workflow automation
• Supabase for transcript and metadata storage
• Windmill for Python workflow orchestration
• Amazon S3 for recording storage
• Zoom API for recording ingestion [END OF SECTION]

Results

[START OF SECTION] After 2 months of implementation, the platform achieved remarkable results:

• 70% decrease in context pings to leadership
• 40+ hours per month saved in manual meeting review
• 100% adoption within 3 weeks of launch

Now the team gets meeting answers in seconds. No downloads, no chasing links, no second-guessing, and no more asking leadership for the same context twice.

The platform has become the go-to knowledge hub for leadership, product, and operations teams—delivering on the original promise of turning every Zoom meeting into a trusted, AI-searchable source of truth. [END OF SECTION]

Goals Achieved

[START OF SECTION] The project successfully met all three original objectives:

1. Ended the cycle of Slack ping-pong — Repeated questions and buried meeting notes are now a thing of the past
2. Transformed raw Zoom recordings — Meetings are now trusted, searchable knowledge without relying on external tools
3. Made leadership context accessible — Information is available instantly without slowing down executives [END OF SECTION]