DX1 / INDIVIDUAL CHALLENGE RISE.DX1.COM.AU

RISE

THE RISE OF AGENTIC SYSTEMS

A solo build. A real problem. A 5-minute demo. Three weeks to ship.

SUBMISSIONS CLOSE

28DAYS · 14HRS · 22MIN · 09SEC

02 · EVENT DETAILS

The challenge.

Pick a real problem. Customer or internal. Build an AI agent that solves it, solo, in three weeks, judged on what it actually does.

100% Agent autonomy

21days To design and ship

5–7min Demo video

$3K In tech prizes

THE RULES

01 Build solo.

One engineer, three weeks, one system. No co-builders, no ghost contributors.

02 Open to all DX1 staff.

Any location, any practice.

03 Pick a real problem.

A customer-shaped problem or an internal DX1 one. Real use cases only, no toy demos.

04 Any model, any framework.

OpenAI, Anthropic, Bedrock, open source. Your call. Build the harness around it: context in, memory across turns, tools the agent can actually call.

05 Show genuine autonomy.

The system reasons and acts on its own. Not assisted prompting wrapped in a UI.

06 Demo end to end.

Walk through what it reads, what it remembers, the tools it calls, and what it actually changed in the world. A non-technical viewer should be able to follow it.

07 Ship a 5 to 7 minute video.

Whatever the framework, whatever the stack, you need to be able to demonstrate it working.

KEY DATES

PHASE 01 / OPENS MON 11 MAY Registrations open. Start building.

PHASE 02 / DEADLINE MON 1 JUN Submissions close 23:59 AEST.

PHASE 03 / WINNERS FRI 5 JUN Announced at the all-hands.

Judging window: 2–4 June 2026.

PRIZES

Winner $1,500

2nd place $1,000

3rd place $500

Paid out as tech of your choice. Laptop, peripherals, dev hardware, you pick. Plus: featured at the next all-hands, written up internally, demo reused in client conversations where relevant.

JUDGING

Reviewed by DX1 practice leads across the four engineering practices, with one external judge from the agentic AI space.

Real problem.

A genuine customer-shaped problem, not a toy demo.

Genuine autonomy.

The agent decides and acts. It is not a prompt wrapper.

Technical depth.

Architecture, controls, evaluation, failure modes.

Demo quality.

A non-technical viewer can follow what happened.

Use whatever stack ships the fastest. Any model, any framework, any cloud is fair game. The agent is judged on what it does, not what's underneath it. One boundary: don't ship anything covered by a client NDA. Questions go to the #rise channel on Teams.

03 · REGISTER

Register.

One paragraph. We'll come back within 48 hours.