Skip to content

Connectors / In development

Dowelify inside the AI you already use.

The connector is being built so a Job can continue inside hosts such as Claude and ChatGPT without turning the conversation into a separate, disposable record.

Different AI, same Job. Next week, next person, next model—nobody starts over.

In development
The mated Dowelify assembly from the finished film: physical work and an AI representation registered by one durable Job reference.

One durable reference

The Job remains with Dowelify.

The planned connector retrieves only the authorized slice needed for the question and records deliberate contributions back into the same Job. A host conversation does not become a second source of truth.

One authority boundary

The model helps. People sign off.

The design keeps the same human boundary across hosts: models can reason and propose, while records of performed, accepted, and released work belong to authorized people.

Scoped context

Show the useful slice, not the archive.

The planned flow provides the model with the Job material needed for the request. Raw media is included only when a person explicitly points the operation at it.

Open connection

Built around MCP.

The connector work uses the Model Context Protocol so Dowelify can expose one governed Job interface to compatible AI hosts. Installation is not publicly available yet.