Test Flight is a two-day, in-person AI hackathon hosted by Glasswing Ventures at 275 Newbury Street in Boston on September 26 and 27, 2026. One weekend, and $300,000 in compute credits on the line for the team that wins.

The brief is Enterprise-Ready AI. Teams build AI-native solutions to problems enterprises actually have, not AI features bolted onto existing workflows. What gets rewarded is a working product demoed live, with a credible path to deployment and a measurable return.

Compute and engineering support are provided throughout the weekend, backed by AWS, Microsoft Azure, and TensorWave. Meals are covered both days.

Schedule. Saturday: registration and breakfast at 9:30 AM, welcome remarks at 10:00, team formation and compute setup at 10:30, building from 11:30 AM until 8:00 PM. Sunday: doors at 9:30 AM, hard code freeze at 2:00 PM, team presentations from 2:00 to 4:30, awards ceremony 4:30 to 5:30.

Who should apply: technical builders, individually or in teams of up to four. Solo applicants and partial teams are matched with other qualified builders. Admission is by application and reviewed on a rolling basis, so applying early matters. Applications close September 19, 2026. Free to participate.

Apply at https://luma.com/gzpv85yq

Glasswing Ventures is a Boston-based venture firm with over $450M in AUM, investing in AI and frontier-technology companies at the earliest stages.

Requirements

What to build: an AI-native solution to a real enterprise problem. The AI should be load-bearing, not decorative. Judges want something that works, not a slide about something that could work.

What to submit, before the 2:00 PM code freeze on Sunday: a working demo you can run live; a short statement of the customer problem and who would buy the fix; the business case, meaning what outcome improves and how a buyer would measure it; your operational assumptions around data, security, and integration; and a link to your repo or deployed project.

Every team presents live to the judging panel between 2:00 and 4:30 PM on Sunday. Judging is on four criteria: Customer Problem, Business Case, Operational Fit, and AI Differentiation.

Hackathon Sponsors

Prizes

3 non-cash prizes

Grand Prize

1 winner

$300K in compute credits, awarded to the winning team. Compute provided by our partners.

Second Place

1 winner

$40K of compute credits

Third Place

1 winner

$15K of compute credits

Judges

Rudina
Glasswing Ventures

Judging Criteria

Customer Problem
Does the project solve a meaningful business challenge that an enterprise would pay to fix?
Business Case
Is there a clear rationale for adoption, with outcomes a buyer could measure?
Operational Fit
Could a real enterprise integrate and run this, accounting for data, security, and deployment constraints?
AI Differentiation
Is the AI load-bearing rather than decorative, and is there a data or model advantage behind it?