Video Knowledge Platform
A video knowledge platform transforms raw video footage into structured, searchable organizational knowledge — making your video library as queryable as a document database. VidForgeX is the enterprise video knowledge platform: it ingests any video, extracts AI-structured insights, builds a semantic vector index, and makes everything accessible through natural language chat and AI agents.
What makes video a knowledge problem
Documents, spreadsheets, and databases have been "knowledge" for decades — searchable, analyzable, and queryable by machines. Video has not. A 2-hour training session, a month of CCTV footage, or a year of workplace recordings contains enormous organizational knowledge — but it's locked inside an MP4 file that only humans can interpret, one frame at a time.
A video knowledge platform solves this by treating AI video analysis as a knowledge extraction step, not just a viewing enhancement. The goal is not better playback — it's converting the intelligence inside footage into structured data that machines can query and AI agents can reason about.
Video knowledge vs document knowledge
Organizations already manage document knowledge in Notion, Confluence, SharePoint, or Google Drive. These platforms are excellent for text — but they cannot extract or represent the knowledge class that only exists in video:
| Knowledge type | Document platforms | VidForgeX video knowledge |
|---|---|---|
| Process documentation | Written SOPs, flowcharts | Observed workflow sequences extracted from actual footage |
| Performance records | Manually entered stats and reports | AI-measured metrics from video (cycle time, sprint rate, utilization) |
| Incident records | Written incident reports (after the fact) | AI-detected events with timestamp evidence from CCTV/operations footage |
| Person history | HR records, manually written profiles | Session-by-session performance and activity timeline from video analysis |
| Search granularity | Section / paragraph level | Activity / timestamp level — "find the 4-second moment where X happened" |
Video knowledge domains
Every match and training session becomes a searchable performance record. Coaches can query "how has player 7's sprint rate trended over the last 12 sessions?" and get a data-grounded answer from the video knowledge base — not from manually entered stats. The knowledge compounds: as more sessions are analyzed, trend detection becomes richer.
Workplace video becomes a continuous SOP compliance audit. Operations managers can query "which workstation has the highest non-value-added rate this quarter?" and receive a ranked, cited answer from cycle-time and activity chunks extracted across dozens of shift recordings. Knowledge that previously required manual time studies now accumulates automatically.
CCTV archives become a searchable investigation database. Security teams can query "find all timestamps where person X was on floor 3 between Oct 1–15" and receive a chronological sighting timeline with camera source and timestamp — drawn from person detection chunks across the surveillance knowledge base.
Making video knowledge accessible to AI agents
The VidForgeX MCP (Model Context Protocol) server at mcp.vidforgex.com exposes 11 tools that any MCP-compatible AI agent can call. This makes VidForgeX video knowledge natively accessible to Claude, custom LLM agents, and automated workflows:
ask_about_session— query the knowledge base for a single video sessionask_about_person— query all knowledge about a specific person across their full video historycompare_profiles— comparative analysis of 2–10 persons from the knowledge baseget_person_progression— metric timeline for a person across up to 50 sessionsget_team_summary— aggregated knowledge summary for a configurable period (1–365 days)get_sop_compliance— compliance knowledge from adherence evaluation results
API keys (vfx_ format) are organization-scoped with configurable expiry and scope restrictions.
Knowledge compounding: Unlike a document knowledge base that requires manual updates, VidForgeX video knowledge grows automatically. Every new video upload adds to the searchable knowledge base — requiring zero manual effort from the team beyond the upload itself.