Winning the BGI Sprint does not mean building the biggest or most complex project.

Winning means producing a clear, useful, reviewable artifact that responds strongly to one of the sprint tracks and shows meaningful potential for BGI, OmegaClaw, ASI: Create, or the broader SNET ecosystem.

One finalist will be selected from each track. These finalists will present their work live to BGI researchers, ecosystem leaders, and invited thought-leaders.

The prize is not monetary. The main award is visibility, feedback, recognition, and the chance for your work to be seen by people who may help it continue beyond the sprint.

Your project should clearly answer the core question of your selected track.

A strong submission is not just “something we built during the weekend.” It should be obvious why the artifact belongs in the track you selected.

Track 1 — Holding the Thread

Your submission should help explain, test, improve, or demonstrate how agents can maintain goals across time, interruptions, and multi-step tasks.

Strong directions include:

• Improving memory or verification in OmegaClaw

• Creating a debugger or visualizer for agent state

• Building a benchmark for thread-holding

• Creating a red-team taxonomy for failures

• Writing a focused research note on working memory coherence

Track 2 — Knowing What You’re Doing

Your submission should show what an OmegaClaw personal agent inside ASI: Create can do that a stateless LLM assistant cannot.

Strong directions include:

• A useful personal-agent workflow

• A self-modeling contribution

• A user-state inspection interface

• A vertical application prototype

• A product workflow that real users could understand and test

Track 3 — Many Goals, Many Voices

Your submission should help groups, DAOs, teams, or communities coordinate when there are multiple goals and real disagreement.

Strong directions include:

• A multi-agent coordination protocol

• A DAO coordination tool

• A benchmark for collective-deliberation agents

• A research note on representing disagreement

• A prototype for surfacing dissent or tracking commitments

Every team must submit a concrete artifact.

Discussion, brainstorming, and concept exploration are useful during the sprint, but they are not enough as a final submission.

Your artifact should be something a judge can open, read, run, watch, review, or use in under ten minutes.

Valid artifacts include:

• Code repository

• Working prototype

• Benchmark

• Research note

• Technical framework

• Workflow demo

• Governance checklist

• DAO coordination tool

• Red-team taxonomy

• Video walkthrough

• Educational material

• Product concept with a usable demo or clear walkthrough

A strong artifact does not need to solve the entire problem. It needs to make one useful contribution clearly.

The most common mistake in a sprint is trying to build too much.

A narrow, complete artifact is stronger than an ambitious but unfinished one.

Good scope choices:

• A small working OmegaClaw extension instead of a full architecture redesign

• A focused benchmark with a few meaningful test cases instead of a broad evaluation framework that cannot be run

• A clear research note on one operational question instead of a general essay about BGI

• A working personal-agent workflow instead of a vague product idea

• A DAO coordination prototype for one real use case instead of a universal governance platform

Before building, ask:

Can we complete, explain, and demonstrate this by Sunday?

If the answer is no, reduce the scope.

Judges will have limited time. Your artifact should be easy to understand quickly.

Your submission should make clear:

• What you built

• Why it matters

• Which track it belongs to

• Who it is for

• How to review it

• What is complete

• What should happen next

Do not make judges guess what your artifact is or why it matters.

For code submissions, include:

• A short README

• Setup instructions

• Demo instructions

• Known limitations

• What part was completed during the sprint

For research or framework submissions, include:

• A clear title

• The main question

• The core argument

• Practical implications

• How the work could be used or continued

For video or walkthrough submissions, include:

• A short description

• A direct link

• What the viewer should pay attention to

• What artifact the video demonstrates

A winning submission should be useful to someone.

That “someone” may be:

• OmegaClaw maintainers

• ASI: Create users

• ASI: Create product team

• BGI researchers

• DAO operators

• SNET ecosystem projects

• Future HyperSprint participants

• Developers building agent systems

• Researchers studying memory, coordination, or beneficial AI

Your team should be able to answer:

Who would use this, evaluate this, cite this, test this, or build on this?

If there is no clear user or consumer, refine the artifact.

Your artifact should connect to Beneficial General Intelligence in a concrete way.

Avoid vague claims like:

• “This helps AI become better.”

• “This supports beneficial AI.”

• “This is good for the ecosystem.”

Be specific.

Examples of stronger explanations:

• “This benchmark helps test whether an agent maintains goals after interruption.”

• “This workflow shows how a personal agent can support continuity across recurring meetings.”

• “This DAO tool helps surface disagreement before a decision is finalized.”

• “This taxonomy gives developers reproducible examples of thread-holding failures.”

• “This interface helps users inspect and correct what their personal agent believes about them.”

Judges should understand how your work contributes to the sprint’s core challenge.

On Sunday, each team will present in a track-level Zoom breakout room. Teams may pitch live or use a recorded video.

Recommended pitch format:

• 3 minutes: pitch or recorded demo

• 2 minutes: judge questions or clarification

Your pitch should answer:

1. What problem did you address?

2. Which track did you participate in?

3. What artifact did you produce?

4. Who is the intended user or consumer?

5. How can someone review, run, watch, or use it in under ten minutes?

6. Why does this matter for BGI, OmegaClaw, ASI: Create, or the SNET ecosystem?

7. What should happen next?

Do not spend most of your pitch on background. Focus on the artifact.

A strong pitch usually follows this structure:

• Problem: what issue you addressed

• Artifact: what you produced

• Demo or evidence: how it works or what it shows

• Value: why it matters

• Continuation: what should happen next

Judges will evaluate teams during the track-level presentations.

They will consider:

• Alignment with the selected track

• Usefulness of the artifact

• Clarity

• Completeness

• Technical or conceptual quality

• Beneficial impact

• Continuation potential

• Presentation clarity

• Whether the team produced a concrete artifact, not only a discussion summary

The strongest submissions usually perform well across most of these criteria. They do not need to be perfect, but they should be clear, focused, and usable.

The sprint does not end when the weekend ends.

All submitted artifacts may be preserved as community resources, and strong artifacts may influence working group cycles, future showcases, or ecosystem development.

To increase your chances of continuation, include:

• What is already complete

• What is still missing

• What support would help

• Who should review it next

• Whether it is relevant to OmegaClaw, ASI: Create, BGI research, DAO coordination, or future HyperSprints

A good continuation note can make your artifact easier to carry forward.

Avoid these mistakes:

• Choosing a track but not addressing its core question

• Submitting only a discussion summary

• Building something too broad to complete

• Forgetting to explain who the artifact is for

• Making the artifact hard to open, run, watch, or review

• Giving a pitch that explains the idea but not the output

• Depending on judge interpretation instead of making your value clear

• Submitting links without access permissions

• Trying to solve all of BGI in one weekend

The goal is not to be exhaustive. The goal is to be concrete, useful, and clear.

Before submitting, confirm that your team can answer “yes” to each question:

• Did we choose one track clearly?

• Does our artifact answer that track’s core question?

• Did we produce something concrete?

• Can someone review it in under ten minutes?

• Is the link accessible?

• Did we explain who the artifact is for?

• Did we explain why it matters?

• Did we keep the scope realistic?

• Can we pitch it clearly in three minutes?

• Did we explain what should happen next?

If the answer to any of these is no, improve that part before submission.

A strong BGI Sprint submission is:

Track-aligned + concrete + useful + clear + reviewable + realistic + connected to BGI.

The best teams will not only show that they had a good idea. They will show that they turned the idea into something others can inspect, use, test, discuss, or continue.