Codemill has announced an AI-powered QC Assistant for Accurate.Video Validate and an Agent-to-Cantemo connection based on the Model Context Protocol. Both products are scheduled to debut at IBC2026 and are aimed at reducing manual exception handling in media supply chains.
The QC Assistant uses a human-in-the-loop model. It analyses media against operational guidelines, flags potential problems and presents recommendations, but leaves final creative and compliance decisions with an operator. Codemill says customer testing reduced a typical 10–15 minute spot check to one or two minutes per asset. The architecture is intended to work with multiple large-language-model providers rather than locking the workflow to one model.
The Cantemo integration exposes governed media operations through MCP so external agents or bespoke chat interfaces can discover assets and answer operational questions. Examples include locating missing media for a project or checking delivery deadlines. This is less about generating content than connecting an AI agent to existing asset metadata and workflow state.
The published performance figures are Codemill's own and should be validated against each organisation's content, error profile and acceptance policy. The announcement also leaves open the permission model, exact MCP tool surface, deployment boundaries and safeguards against unintended changes to production assets.
