← Projects

Website & AI agents

Bands in Bend

Bands in Bend is a live music guide for Bend, Oregon. I built the backend, the owner’s review app, and the instructions and tools its AI agents use to find and organize show listings.

Visit Bands in Bend
Bands in Bend’s public website, with this week’s live music guide and the owner’s selected show highlights.
The public website · October 2026. Visual design by JDM Creative.

Finding the shows

Many venues in Bend don’t keep their website calendars up to date. A show might get a single Instagram post or a story, and that’s it. Finding out what’s happening means checking a lot of places.

Bands in Bend’s agents do that research using the instructions and tools I gave them. They gather show details from venue websites and social posts, keep links to their sources, and bring the findings into one place. They can also discover new venues and shows to add to the owner’s review queue.

Reviewing the listings

Each AI-sourced finding includes a confidence score and an explanation. The owner can check the source, correct details, and approve or reject the listing. Conflicting times, possible cancellations, and duplicate shows can be flagged for review.

I connected the approved listings to the public website. The owner also uses the calendar to prepare weekly graphics for promoting the bands and shows.

What the agents learn

The system keeps the agents’ original findings and confidence scores alongside the owner’s review decisions and corrections. Agents can use that history to see what they got wrong and adjust their research on later runs.

They also keep notes about where each venue posts its listings and how often to check. A venue might publish a monthly calendar, for example, while another announces shows individually on Instagram. Agents can suggest changes to the checking schedule for the owner to review. Future runs use the saved notes and schedule instead of starting that research over.

Adding another agent

The owner can onboard another agent with a one-use setup link. Shared instructions and tools tell it where to find sources, how to research shows, and how to submit findings for review. The setup works across different AI apps. Each agent gets its own connection, which the owner can revoke.