User enters startup idea via voice
User enters startup idea via text
User enters startup idea via video
Welcome screen
Puffin report generated based on founder interactions
Puffin report generated based on founder interactions
Default dashboard view
Dark mode
Querying all agents - results in agent context
Tasks and Recent Activity
Agent execution
Completed tasks

💡 Inspiration

We’ve entered an era where solo founders are more common than ever. Yet, they still face skepticism from investors who seek evidence of execution, clarity, and velocity. Meanwhile, tools like ChatGPT, Notion, and Replit help — but they’re fragmented and reactive.

I wanted to build something different:
A true AI cofounder for solo founders and a real-time insight platform for investors evaluating them.

🚀 What It Does

Puffin is a cofounder-in-a-box for solo startup founders and a signal amplifier for VCs.

For Founders:

Puffin turns an idea (via text, voice, or video) into a structured startup plan through autonomous AI agents:

📦 Business model, GTM strategy, and ICP
💻 MVP architecture and tech stack
🎯 Pitch deck content and investor communications
💸 Pricing plans and early financial projections

For Investors:

Puffin provides a founder progress intelligence layer:

✅ Track real-time outputs from founder–AI interactions
📈 Monitor weekly “founder velocity” and milestone completion
💬 Receive automated founder updates and pitch iterations
🔎 Analyze how founders respond to feedback or market shifts (via agent diffs and interactions)

This allows VCs to confidently evaluate, back, and support solo builders from day one.

🛠️ How I Built It

Frontend: React + Tailwind with Puffin branding (orange theme, modern UI, dark/light toggle)
Multi-Agent Engine: Toolhouse to define autonomous agents for each business domain
LLM Engine: Groq calling llama3-70b-8192 for ultra-fast LLM completions per agent
VC Layer: Backend feed and optional login for investors to view agent outputs, founder inputs, and logs
Memory System: Timestamped logging of all founder inputs and agent responses for tracking, iteration, and reporting

🧗 Challenges I Ran Into

Token limits across LLM systems
Architecting thoughtful agent-to-agent collaboration
Ensuring agent outputs stay explainable and editable for founders
Turning founder–agent interaction metadata into actionable investor insights

🏆 Accomplishments I'm Proud Of

🤝 Reduced friction for solo founders to get funded and for investors to make confident decisions
🔥 Created a fully functioning AI command center with live agent updates
📊 Built a prototype investor dashboard showing founder–AI interactions and startup health signals

📚 What I Learned

LLM agents are more effective when framed as cofounders, not just tools.
VCs are eager to fund solo founders when there's clarity and velocity — Puffin makes that visible.
The value lies in the feedback loop:
Founder inputs → Agent outputs → Tracked improvements → Investor insight

🔮 What’s Next for Puffin

💼 Launch full-featured Investor Dashboards for funds backing solo founders
🧭 Release Puffin Signals — a “founder velocity score” based on engagement and agent feedback loops
🤝 Build a community to connect solo founders with AI-comfortable investors
🔒 Add permission controls for investor subscriptions and private updates
🌎 Expand globally to support solo founders in emerging markets