Buildathon A2Z

Inspiration

Running Builder Pulse for AABW's own Builder Experience track showed us the pain from the inside: organizers spend weeks stitching together registration pages, sponsorship decks, judge outreach, submission tracking, and reporting across a dozen disconnected tools before a single builder writes a line of code. GenAI Fund alone runs this cycle across multiple programs and countries every year. That repeated, avoidable overhead is what Buildathon A2Z (AI EventOps) is built to remove.

What it does

Buildathon A2Z turns a natural-language event request into an operational innovation program. Given a short description of the event (type, size, timeline, goals), it generates the event structure and timeline, drafts landing and registration page copy, produces sponsorship proposal and outreach materials, drafts marketing content, recommends judges, mentors, and speakers from a structured sourcing pass, and sets up a submission and evaluation workflow, cutting a multi-week launch cycle down to hours.

How we built it

The platform extends the organizer-facing foundation and hackathon domain model we built for Builder Pulse, adding new agent modules for the parts of the event lifecycle Builder Pulse doesn't cover: content generation for landing pages, sponsorship decks, and marketing copy; outreach sourcing for judges, mentors, and speakers; and submission and evaluation workflow setup. Each module is a distinct agent coordinating through a shared event-state model, so the system can go from a single prompt to a structured, multi-artifact program plan.

Challenges we ran into

Balancing genuine automation against usefulness. Generic AI-written sponsorship decks or marketing copy are worse than no output at all, so we had to ground every generated artifact in real structured inputs like event goals, audience, and timeline rather than filler text. Keeping the system honest about its own scope, rather than promising a one-click full event in a box, was an explicit design constraint throughout.

Accomplishments that we're proud of

Reusing real organizer pain we experienced firsthand with Builder Pulse to scope this build, instead of guessing at what event organizers need.

What we learned

Event operations is a genuinely multi-agent problem. Planning, content generation, outreach, and evaluation each need different grounding data and different failure modes, and forcing them into one monolithic agent produces worse output than coordinating specialized agents.

What's next for Buildathon A2Z

Deepening the judge, mentor, and speaker sourcing module with real network data, and piloting the platform on an upcoming GenAI Fund program.

Built With

• node.js
• openai