The Tireless Hand Challenge

A one-day hackathon to build an AI agent that tests software the way a real user would.

Build The Hand That Never Gets Tired

One Saturday, one hard problem, and a room of people trying to crack it from completely different angles.
You get the problem statement in the morning, seven and a half hours to build, and a demo slot in front of everyone that evening.
Bring whatever stack you trust: Playwright, Selenium, browser-use, MCP, or something you’ve been meaning to try. The approach is entirely yours.

Offline @ FlytBase Labs Online

Date

26 SeptemberSaturday, 2026

Time

9 AM – 7 PMIST

Venue

FlytBase Labsor Online

The Challenge

Someone on your team still re-clicks the same screens to check nothing broke. Build the agent that does it instead: drives a real app on its own, heals itself when the UI shifts, and knows a real bug from an intended change. No brittle selectors, no test suite to babysit.

Automated browser testing agent navigating a UI

The Room You Build In

A day with some of the sharpest AI-native builders you’ll meet: engineers, QA leads and researchers, all attacking the same problem from different angles. Ideas get traded openly here, and the people you meet tend to stick around long after.

A wall of monitors rendered in code, tracking runs across an application

Leave With Something Real

You walk out with an agent you can point at your own codebase on Monday, plus an honest read on how far it actually goes. The FlytBase engineers live in this problem every day, so their feedback is usually what you value most.

Who Gets Into The Room?

Run a scan to find out.

output.log

01 · CURIOSITY COUNTS

Curiosity Counts

Offline vs Online

Offline · FlytBase Labs

Build alongside a curated group of engineers, QA automation experts, and AI/LLM builders.
A session on the current state of the art and live demos from the FlytBase team.
Real-time feedback as your approach takes shape.
Breakfast, lunch, and snacks on us.

Online · Open To Everyone

Same problem statement, released at the same moment.
Same build window, same evening demo slot.
Same evaluation, same shot at getting noticed.

Structure Of The Day

9:00 AM

Breakfast + Introductions

Start with breakfast, coffee, and the people you'll be building alongside throughout the day.

9:30 AM

State Of The Art + Problem Statement

A look at where autonomous UI testing stands today, demos from the FlytBase team, a walkthrough of the Tireless Hand Challenge problem statement, and Q&A.

10:30 AM

The Hackathon Begins

Get seven and a half focused hours to build. Ship one reliable, self-healing testing agent end to end, or go deep on the hardest part of it.

6:00 PM

Selected Demos + Results

See a range of technical approaches from across the day. Selected builders share what they built, the results they got, and what they learned.

Reliability comes first. Everything else is secondary until the system can consistently complete flows without breaking. Full scoring breakdown below.

Evaluation & Scoring

What Judges Are Looking For

Reliability & Self-Healing: does it keep working when the app changes, a selector shifts, or a flow is reordered? It should recover and adapt on its own rather than failing and waiting for a human fix.
Context & Decision Making: can the system ingest business rules and accurately distinguish between an actual bug and an intended UI/feature update?
Cost & Efficiency: does the solution optimize its reasoning and actions? Solutions that execute regression suites reliably without massive API/token costs on every single step will score highly.
State Memory (Self-Improving): solutions that learn the app's structure and avoid rediscovering the same controls from scratch on subsequent runs will be rewarded.
Test Generation & Maintenance: can the tool automatically generate a test suite for an app that has none, and automatically deprecate or update those test cases as the app evolves?