Intent-Driven Hackathon (Claude Code Edition)
- Mark Kendall
- 2 days ago
- 3 min read
Instead of organizing the hackathon like a normal coding sprint, you structure it around Intent → AI → Implementation → Demonstration.
Let’s design it so people experience the productivity jump, not just hear about it.
Intent-Driven Hackathon (Claude Code Edition)
Goal
Prove that small teams using intent-driven architecture + Claude Code can build something meaningful in 1–2 days that normally takes weeks.
Your demo success metric:
“We built a working intelligent system from idea → architecture → running software in a single hackathon.”
Not just code — a real system.
1. Team Structure (The Secret Sauce)
Instead of normal dev teams, use intent-driven roles.
Each team = 4–5 people
1️⃣ Intent Architect (Team Lead)
This is the most important role.
Responsibilities:
Define the intent.md
Clarify what the system should do
Keep the AI aligned
Refine prompts
Think:
Product Owner + Architect + Prompt Engineer
2️⃣ Claude Operator
This person drives Claude Code inside VS Code.
Responsibilities:
Running Claude commands
Generating architecture
Generating code
Running builds
Fixing errors with AI
They are basically the AI pilot.
3️⃣ System Integrator
Handles:
running services
docker
APIs
debugging
wiring components together
They keep the system actually running.
4️⃣ Domain Thinker
Keeps the solution useful and realistic.
They answer:
What problem are we solving?
What would a real user do?
What data do we need?
5️⃣ Demo Builder (Optional)
Focuses on:
UI
Demo scripts
Slides
storytelling
Because hackathons are won by demos.
2. The Hackathon Theme
You want something AI-native, not just CRUD apps.
Good hackathon challenges:
Option A — Autonomous Architecture Analyzer
Input:
GitHub repo
Output:
architecture summary
dependency map
technical debt report
suggested improvements
Using Claude Code.
This one fits your consulting world perfectly.
Option B — Enterprise Intent Generator
Given:
a repo
or requirements
Generate:
Then generate the system automatically.
Option C — AI Operations Assistant
Input:
logs or telemetry
Output:
root cause analysis
remediation plan
automated fixes
This one impresses leadership.
Option D — Developer Onboarding Agent
Give it:
a large codebase
Output:
What this repo does
How to run it
Where the APIs are
Architecture overview
Basically an AI onboarding assistant.
Consulting gold.
3. Hackathon Timeline (1–2 Days)
Hour 0 – Kickoff (30 minutes)
You explain:
Intent → Claude → Code → System
And the rules:
Everything starts with intent.md
Claude must generate most code
Teams show Claude interaction history
System must run
Hour 1 – Intent Creation
Teams write:
Example:
System Intent
Build an AI system that analyzes a GitHub repository
and generates an architectural summary including:
- service boundaries
- dependencies
- technology stack
- potential improvements
Then Claude generates:
Hour 2 – Architecture Generation
Claude generates:
services
APIs
repo structure
tech stack
Example output:
services/
repo-analyzer
architecture-generator
report-generator
ui-dashboard
Hours 3–6 – Build Phase
Teams:
run Claude Code
generate services
test endpoints
refine prompts
You will see insane velocity here.
Hours 6–8 – Demo Prep
Teams prepare:
1. The Intent
2. The Claude interaction
3. The Architecture
4. The Running System
4. Judging Criteria
This is important.
Not just code quality.
Score based on:
Category
Description
Intent Quality
Was the intent clear and powerful
AI Utilization
Did they use Claude effectively
Architecture
Clean system design
Innovation
Is this something real
Demo
How impressive is the demo
5. The Hackathon Twist
Add this rule:
Teams are allowed to change the intent mid-stream.
This shows the power of intent-driven development.
Example:
Claude, update the system to support multiple repos.
Watch the architecture evolve.
That’s mind blowing in demos.
6. The Final Demo Format
Each team gets 7 minutes.
Structure:
1️⃣ The Intent
Here was our intent
2️⃣ The Claude Conversation
Show prompts like:
Claude, generate the architecture
Claude, implement the repo analyzer service
Claude, create the API
3️⃣ The System
Live demo:
Upload repo
Analyze repo
Generate architecture report
4️⃣ The Result
Example:
This system took 6 hours to build.
Normally it would take weeks.
That sells the idea.
7. Why This Hackathon Will Work
Because you are demonstrating something bigger:
Not just AI coding.
You are demonstrating:
Intent → AI → System Creation
Which is basically the future of engineering.
8. The Big Reveal (Your Moment)
At the end you close with something like:
“What you saw today isn’t just AI coding.
It’s the beginning of intent-driven engineering.”
Boom.
That’s how movements start.
Comments