AI OMEGA AGENT DEV BUILDATHON

AI Omega Lisbon Developer Buildathon — Rules & Brief

Two days, one working Omega agent, built against a real problem from the XRAGENCY portfolio.

23–24 October 2026 · Lisbon in person, and online from where ever you are

At a glance

When	Fri 23 and Sat 24 October, 10:30–17:00 Lisbon time
Venue	HealthHub, by Unicorn Factory Lisboa
Seats	20 in person. Approval required — registering does not secure a seat
Teams	Solo, or up to four
Submit by	13:00 Saturday 24 October. No late submissions
Prizes	No cash. Selected teams pitch to the CEO of Superintelligence

Yes, use AI. Claude Code, Cursor and other coding assistants are welcome. Credit them in your README.

The idea

You build an Omega agent that solves a real problem for a real organisation. Every track comes from a live project in the XRAGENCY portfolio, so your work has a user from day one.

Omega is SingularityNET's neural-symbolic agent framework, built on the Hyperon stack. It pairs a large language model with a MeTTa reasoning layer, keeps long-term memory, and leaves an auditable trail of why it acted.

Your build must show three things: the agent's role in the workflow, what it remembers or reasons about, and the tools it calls.

Past winners, for the bar: Mizani (two agents reconcile a maternal health referral, reasoning in Swahili), HelixMind (questions to evidence-backed research answers), Venture Route (MeTTa rules match a founder's brief to verified builders).

Entry types

	New build	Existing project
Goal	Idea to working prototype	Prototype to something closer to production
Must show	A working flow by the showcase	Your baseline, and exactly what you added on 23–24 Oct

State your entry type and track at the top of your README. Existing projects: tag your baseline commit in Git.

The tracks

Pick one primary track. Switching is allowed until 13:00 on day one.

Track 01 — Rejuve.io, Biomedical research assistance

• Problem — research teams need traceable answers across scattered biomedical evidence.
• Build — an Omega research assistant linking a focused question to relevant public research and structured evidence.
• Show — a source-linked answer with its reasoning and uncertainty, plus a test where evidence conflicts or is missing.
• Boundary — public, synthetic or explicitly authorised data only. Research support, not medical advice.

Track 02 — Omega payouts, Global payouts and prize transparency

Stellar testnet · Raven

• Problem — BASIX needs transparent prize funds and reliable payouts across continents.
• Build — an Omega payout coordinator and a public prize-fund ledger on Stellar testnet. Use Raven for Stellar support. Simulate any NEAR Intents cross-chain leg (it has no testnet).
• Show — a published funding rule, approval before any transfer, each payout linked to a receipt, failed or duplicate requests handled, reserve carried to the next event.
• Boundary — testnet and dry quotes only. No live funds.

Example rule. 20 eligible teams of a 50-team target releases 40%. A $1,700 fund gives $680 for awards and $1,020 reserved for the next event, before fees.

Track 03 — BASIX.MARKET, Builder and marketplace workflows

• Problem — builders need a clearer route from verified skills to suitable projects.
• Build — an Omega agent for one workflow: project discovery, contributor matching, onboarding or submission review.
• Show — a recommendation with its evidence, and how it handles an incomplete or unsuitable profile.
• Boundary — sample profiles and briefs, or authorised access. Name any mocked services.

Track 04 — World Tennis, Player and club coordination

• Problem — players and clubs coordinate matches, availability and bookings across disconnected tools.
• Build — an Omega assistant for one focused workflow, such as matching available players to a suitable court and time.
• Show — a proposed match with reasons, a resolved scheduling conflict, and a confirmation step before a simulated booking.
• Boundary — sample player profiles and court availability. Label simulated bookings and payments.

Before you arrive

Setup done by 10:30 on Friday. Teams that install in the morning lose half of day one. There is a setup clinic at 10:30 if you are stuck.

• Laptop with Docker working, and Omega running in it — github.com/singnet/Omega
• An LLM API key Omega supports — Anthropic, OpenAI, ASI Cloud, ASI:One or OpenRouter
• A channel to talk to your agent — Telegram, Slack, Mattermost or IRC
• A GitHub account, an empty repository, and your coding assistant set up
• A BASIX.MARKET account — basix.market
• Track 02: Stellar testnet account with Raven connected
• Existing projects: baseline commit tagged

Schedule — Lisbon time

	Fri 23 — build	Sat 24 — showcase
10:30	Check-in, seat approval, setup clinic	Doors open, building continues
11:00	Welcome, rules, track briefs	Mentor office hours (online)
11:45	Building starts	Building
13:00	Lunch	Submission deadline, then lunch
14:00	Mentor office hours (online)	Running order confirmed
14:15	Building	Filmed showcases, 3 min per team
16:30	Progress check: each team shows what runs	Wrap-up and next steps
17:00	Close	Close

Joining remotely

Registrants who aren't in Lisbon in person take part fully online — same tracks, rules, deadline and selection criteria. The 20 seats are in-person only; approved remote participants get the link before Friday.

Remote-only and mixed teams are both fine. Same setup checklist, plus a stable connection, webcam and microphone. Present live in the running order and keep a recorded backup. Remote showcases are filmed with consent and eligible for the pitch.

The rules

1. One primary track. Name the track and the user problem at the top of your submission.
2. Omega must do real work. Show its role, memory or reasoning, and the tools it calls. A chatbot wrapper does not count.
3. Mark your new work. Existing projects are welcome. Show the baseline and what you added on 23–24 October.
4. Every attendee needs approval. Each team member, individually. Registering does not secure a seat.
5. Credit everything. Each contributor, reused code, datasets and AI assistance. Respect licences.
6. Run Omega in isolation. It can run shell commands and change files. Use Docker, never a machine with real credentials.
7. Authorised data only. Public, synthetic, sample or explicitly authorised. Never personal health data or real client records.
8. No real money. Payment features run on testnet or in simulation, and say so.
9. Protect credentials. Keep API keys out of your repository and off your demo screen.
10. Behave professionally. Harassment, cheating, plagiarism or misrepresenting your work means removal.

Submission package

Due 13:00 Saturday 24 October, through the basix.market/lms/buildathon/4/ entry submissions page. What should be submitted:

1. Working prototype — runnable, or an accessible demo. Label every mocked component.
2. README — track, entry type, problem, setup steps, dependencies, contributors, and new work since your baseline on the GitHub repo shared with you at basix.market website buildathon registration.
3. Three-minute demo — the user flow, what Omega contributes, the result.
4. Evidence and disclosures — test results, data sources, AI tools used, known limitations.

If someone else cannot run it from your README, it is not finished.

Showcase and selection

Three minutes per team, filmed by beyondthecode.ai: the problem, a working demonstration, your contribution, your next step.

The review panel chooses who pitches to the CEO of Superintelligence on:

Weight	Criterion
30%	Working execution — a useful flow that works, backed by test evidence
25%	Track fit — a clear user problem and a credible benefit for that organisation
25%	Omega integration — a meaningful agent role the team can explain
20%	Demo and docs — a clear story and enough detail to reproduce the work

Ownership, data and consent

You own what you build. Taking part transfers nothing to BASIX.MARKET, XRAGENCY or any track organisation; commercial terms are agreed separately, in writing. Track organisations own their data and brands — use their names, logos and materials only as the track brief shows. beyondthecode.ai asks before publishing identifiable footage, and you can decline filming and still take part.

After the build

No cash prizes. You leave with a filmed showcase of your build, a feature on BASIX.MARKET and beyondthecode.ai for selected projects, and — for selected teams — a pitch to the CEO of Superintelligence. Outcomes are shared with everyone afterwards, and strong builds can lead to follow-up work with the project they were built for.

FAQ

Do I need Omega or MeTTa experience? No. Come ready to learn fast; mentors are online both days.

Can I come alone? Yes. Build solo, or form a team of up to four on day one.

Still stuck? office@basix.market

Links

• Omega source and setup — github.com/singnet/Omega
• Raven and Stellar development — developers.stellar.org
• Registration and participant access — basix.market
• Video showcase — beyondthecode.ai
• Contact — office@basix.market