Before you press play
What you’ll find in the video
- Ask separate Choice questions about the same issue title and description: one for bug, enhancement or documentation, and another for the coding-agent label. Jev chooses labels; the coding agent does the work.
- Use current official TypeSafe instructions, then test a representative issue. Inspect the chosen labels and returned probabilities before connecting them to a workflow that acts on the issue.
- Keep a record of the input and response, with sensitive data kept private. Decide how uncertain routes should be handled: the example still assigns an agent label with a reported confidence of 0.62.
Worth knowing
The suggested 0.75 human-review cutoff is not shown as a validated, enforced gate. The million-classification price comparison is the creator's assumed-input estimate, not a reproduced cost or accuracy benchmark. Agent names, models, account access and prices reflect the recording, not a current recommendation. The demo repo is offered through a free Skool community, not a direct public implementation link, and its workflow was not audited here. Includes Decodo sponsorship and community promotion. These notes use retrieved untimed caption text and selected video evidence, not a full audiovisual watch.