contextforce turns people who know — a creator’s videos, a website, a folder of files — into expert agents. Each question goes to the agent with measured expertise for it, comes back with its sources, and turns into action.
Creators from any platform go through one ingest pipeline — tagged, extracted, assessed for expertise, and kept as knowledge. That is the agents’ brain, and any AI assistant or app can plug into it.
A creator’s whole channel, a website, or a folder of files. contextforce fetches it once and caches it; the agent keeps what it learns as its own knowledge base.
Nothing is shared between agents, so an answer only ever comes from that agent’s sources.
Every video is tagged with where it is and what it’s about. A creator’s level in each place × subject comes from how their videos there perform against their own average — views and engagement — and how deep they go.
That map is what Auto reads when it picks who answers.
A question searches only the agent’s own knowledge and comes back with passages — each tied to the video and the moment it was said, the page, or the file.
Fast enough to sit inside a chat turn or someone else’s assistant.
Chat with it, call its API from an app, or reach it over MCP from Claude, ChatGPT or any assistant — where your own model writes the answer and the agent supplies the sources.
Under every answer, the expert offers what to do with it: build a list, show the route, plan the trip, find flight deals, poll the group, send an RSVP.
Each one is made from the expert’s own picks and lands as something you keep and share.
Keyword search matches words; semantic search matches meaning. Both miss what the shopper needs. contextsearch reads the intent and ranks by what fits — more searches end in a sale.
Every creator the pipeline scores can become an agent; so can a website or a folder of files. Each joins with its own knowledge and its measured expertise, and Auto starts sending it the questions it knows best.