Summary
The core value proposition demonstrated in the available transcript is that RecapHub can turn event recordings into event-specific, reusable content assets by grounding extraction in a custom knowledge graph rather than randomly selecting moments. The discussion opens with how “moments” are generated from prompts, angles, agents, and a knowledge graph tailored to the event and discussion topic. The group also briefly flags a practical production issue: the quality of recordings, the lack of camera operators, and a proposed higher-quality setup costing about 45k, before shifting into how RecapHub’s outputs are actually used after an event.
The product workflow is presented as two layers: the on-platform session clipper and an expanding set of agents that automate desired outputs as soon as a session finishes. The speakers describe integrations such as Google Drive, and possible future or co-developed integrations with Luma, LinkedIn, and Instagram. A key HumanX example shows the intended speed and distribution model: for every session, RecapHub generated clips plus a “tech crunch style article,” then sent the recording, clips, and article to speakers, PR agencies, press, and the marketing team within about twenty minutes after the session ended. A running press room was also created so important session outputs could be pushed to journalists on the ground.
The middle of the transcript focuses on what ProsusAI would need to do to operationalize this. If there is no Luma integration, RecapHub can create blank session placeholders, then process video once it arrives; with integration, session titles, speaker names, roles, companies, and descriptions can be brought in automatically. The team discusses preparing reusable templates and knowledge-graph nodes for technical demos, including upcoming tech takeaways, predictions, research, AI relevance, and technology stacks such as front end and back end. They frame upcoming events as tests to understand sustainability, dependency on the RecapHub team, and how the outputs can be leveraged internally.
The later available section explores distribution beyond LinkedIn. A press-room style discovery page can compile moments from multiple talks into stories, link each story back to the exact video moment, and potentially serve attendees, internal stakeholders, social media managers, or the wider Prosus ecosystem through public links, logins, or embedded pages. The transcript then introduces MCP as a way for clients to access transcripts, analysis, and knowledge graphs to build assets not already available on the platform. As an example, RecapHub used Claude with World Economic Forum sessions from 2025 and 2026 to create an interactive comparative report with quotes and statistics. The provided transcript ends during this comparative report demo.
Key Topics
5 key topics from Saptarshi, Prakhar, Mila, Ashley and Sara at Prosus AI House. Thicker branches were mentioned more often in the talk.
Key Takeaways
- Predefine knowledge-graph prompts before each event.
- Automate clips, articles, and speaker delivery after sessions.
- Use press-room pages to surface cross-session stories.
- Test upcoming events to refine templates and dependencies.
- Expose transcripts through MCP for custom reporting.
“Every event that we do is kind of a hackathon for us.”