Yes. For Devpost, I’d make it tighter and more structured than the Luma email. Below is copy-paste text for the main sections Devpost normally asks you to complete.

Global AI Hackathon — Powered by Mel

Build something useful. Make it agentic. Ship it.

Build and ship an agentic project with Mel, the agentic terminal.

Choose from four tracks: Agentic Commerce, Agentic Finance, Agentic Marketing, or Agentic Engineering: Debug, Test & Ship.

Build solo or in a team of up to 3. No coding experience is required.

Welcome to the Global AI Hackathon — Powered by Mel, presented by Mel × Agentic Day.

This hackathon is about building AI systems that do more than answer prompts.

We want you to build something agentic: a system that can understand a goal, plan, take actions, use tools or data, observe results, adapt, and move a task toward completion.

Think:

Goal → Understand → Plan → Act → Observe → Decide → Act Again → Verify

Your project does not need to be huge or fully autonomous.

A focused agentic workflow that genuinely works is better than an ambitious project that cannot be demonstrated.

Build with Mel throughout the hackathon to plan, code, run commands, research, debug, test, iterate, and verify your project.

Download Mel:
https://www.openmel.dev

Anyone can participate.

You can build:

• Solo

• As a team of up to 3 people

Each participant may be part of one submitted project/team.

Developers, designers, founders, students, AI builders, product people, and first-time hackers are welcome.

No coding experience is required.

Choose ONE of the following four tracks and build a working agentic project.

Your project should:

• Solve a real problem

• Include a working user flow

• Demonstrate meaningful agentic behavior

• Use Mel materially during the build

• Be demonstrable by the submission deadline

The agentic functionality must be part of the working project — not simply described in your presentation.

Build an agentic system that helps businesses sell, operate, support customers, or make better commerce decisions.

• AI shopping or product-discovery agent

• Inventory optimization agent

• Merchandising or promotion agent

• Returns and customer-support agent

• Product bundling or recommendation system

• Agentic storefront

You may start from an existing open-source commerce project, template, or a project you have permission to use.

Use Mel to:

Understand → Run → Modify → Add Agentic Capabilities → Test → Verify

For example, take an existing storefront and turn it into an agentic shopping experience that can understand customer intent, search products, compare options, react to unavailable inventory, and recommend the next action.

If you start from an existing project, clearly identify the original project and show what existed before and what your team added or changed during the hackathon.

Build an agentic system that helps individuals, founders, or organizations understand or operate their finances.

• Cash-flow agent

• Expense-analysis agent

• Financial reporting agent

• Runway or budgeting assistant

• Invoice-management workflow

• Subscription monitoring

• Financial anomaly detection

A strong finance workflow might:

Read Data → Detect → Investigate → Calculate → Recommend → Flag for Action

Use public, mock, or sanitized financial data.

Do not build tools that provide personalized trading or lending advice.

Build an agentic system that can research, create, operate, test, or improve marketing.

• Campaign planning agent

• Audience research agent

• Content workflow

• Landing-page agent

• Lead-research workflow

• CRM workflow

• Campaign-analysis agent

• Marketing experiment engine

Go beyond simply generating content.

A strong workflow might look like:

Objective → Research → Plan → Create → Evaluate → Improve

Your agent should help move a marketing objective toward an actual result.

Build or work on a real software project and use Mel throughout the engineering process.

You may:

• Bring an existing GitHub repository

• Use an open-source project

• Use a broken starter project

• Build a new project containing a meaningful engineering challenge

Use Mel to:

Understand → Investigate → Plan → Debug → Modify → Run → Test → Fix → Verify → Ship

We want to see whether Mel can help you work through a real codebase and solve problems — not simply generate code.

Your demo should clearly show:

• The original problem

• How you approached it with Mel

• What changed

• How it was tested

• Evidence that the final solution works

We are looking for more than:

Prompt → AI Response

A strong submission should demonstrate some form of:

Goal → Understand → Plan → Act → Observe → Decide → Act Again → Verify

Your agent might use APIs, databases, files, tools, search, business logic, external services, or other systems to accomplish its goal.

It does not need to perform every step autonomously.

Working agentic behavior matters more than the number of agents you create.

Mel is the development environment powering this hackathon.

Use Mel during your project for:

• Planning

• Building

• Coding

• Running commands

• Research

• Debugging

• Reviewing

• Testing

• Fixing failures

• Iterating

• Verifying your final project

Download Mel:
https://www.openmel.dev

We also want your feedback.

Tell us what worked, what didn't, where Mel helped, and where you encountered friction.

Each team submits ONE project.

Your submission must include:

1. Project name

2. Selected track

3. Team member names — maximum 3

4. Problem statement — what problem are you solving?

5. Solution — what did you build?

6. Agentic workflow — what makes your project agentic?

7. GitHub repository and/or working project URL

8. Short demo video showing the project working

9. 3 ways Mel helped you

10. 3 issues, failures, or friction points you experienced with Mel

11. 1 feature you would like us to add to Mel

If you use an existing open-source project, template, or codebase, identify the original source and clearly explain what you added or changed during the hackathon.

Your project must be created or meaningfully developed during the hackathon.

You may use:

• Open-source projects

• Existing frameworks

• APIs

• AI models

• Public datasets

• Templates

• Libraries and developer tools

However, your team's hackathon contribution must add meaningful new functionality.

If you begin with an existing project, you must disclose it.

You must have the right or permission to use any code, assets, datasets, APIs, or other materials included in your submission.

The agentic functionality must actually be implemented in the project.

Judges will evaluate what you can demonstrate, not just what you planned to build.

Does it solve a real problem?

We will evaluate the value of the problem, who benefits from the solution, and whether the project could be genuinely useful.

Does the core workflow actually work?

Judges will consider functionality, completeness, quality of execution, and whether the team can demonstrate the project working.

Is the approach interesting and thoughtful?

We are looking for creative applications of agentic AI, not simply another chatbot.

Did Mel materially contribute to building the project?

Show us how you used Mel to plan, build, debug, test, review, iterate, or ship.

A strong submission should answer five questions clearly:

1. What problem are you solving?

2. What does your agent actually do?

3. What actions can it take?

4. Can you demonstrate that it works?

5. How did Mel help you build it?

A smaller working project can score higher than a much larger project that cannot be demonstrated.

https://www.openmel.dev

https://globalhackathon.agenticdaysummit.com/submissions

https://chat.whatsapp.com/F7L3bjLy51a1GTuxP0tI3C

https://discord.gg/YRC92w5A8c

Our mentors and support team are available throughout the hackathon.

Use WhatsApp and Discord to ask questions, get technical support, connect with mentors and other hackers, share progress, and give us feedback about Mel.

If you get stuck, reach out.

Build Window: 36 hours

Sunday Live Session:
11:00 AM–2:00 PM ET

Final Submission Deadline:
Sunday at 2:00 PM ET

During the Sunday live session, mentors and the support team will be available for technical help, project feedback, demo preparation, and submission support.

Keep your scope focused.

You don't need to build the biggest project.

Build something useful. Give your agent a meaningful job. Make the core workflow work. Test it. Show us what works — and what doesn't.

Build something real.
Make it agentic.
Test it.
Verify it.
Ship it. 🚀

Mel × Agentic Day

https://www.openmel.dev

• production ready
  You will be judge on deploying a production ready project.