agentic board members
landing page
agentic intern cookin
disagreeing agents
personalize agents
generate a pitch for investors

Inspiration

Building startups can be overwhelming, so we asked: what if you had your own army of agents ready to guide you? Inspired by the chaotic, creative energy of a real startup team — the Unpaid Intern, the Overworked CTO, the Enthusiastic PM — we built a platform where agents come to life to help you explore infinite possibilities for your next move, be it your next marketing move or personalised business feedback from the CEO himself on how well it'll do.

What it does

Our web app lets you pitch a startup idea and watch a group of personality-driven AI agents be your discussion boardroom. Each agent has a distinct personality, from eager optimism to tough realism. Together, they generate ideas, challenge each other, and strategize in real time. You (the user) can hop in, guide the discussion, or just sit back and watch the magic happen. They will be able to think of things that wouldn't have occured to the user. At the end, you receive a polished pitch deck you can use, a report with the conversation summarizing the best ideas for your next steps, a mood board and much more!

How we built it

We combined several technologies to bring this vision to life: Agent Creation: We built a custom agent factory that spawns a consistent team of agents, each seeded with a unique personality archetype. Agents communicate via event-driven messaging, naturally building debates, collaborations, and even a little drama to keep things engaging. ASI:One Mini Integration: FetchAI’s ASI:One Mini LLM powers the brains behind each agent, guiding their tone, responses, and decision-making processes based on their assigned roles. Human-Agent Communication: Using Flask and REST endpoints, users can inject their own ideas into the conversation, acting as the "boss" while the agent team adapts and responds dynamically. Web App Interface: We built a React frontend to visualize the agent discussions in real-time. Our Flask backend actively pushes agent conversation updates to the frontend, allowing users to intuitively interact with their "agentic boardroom." Final Report Generation: After the session, we pipe all conversation context into Gemini to create a summarized, actionable meeting report, neatly organized for the user to review.

Challenges we ran into

One of our biggest challenges was getting the agents to actually talk to each other the way we envisioned. At first, we figured out how to make agents respond to us, but making them interact with each other, while staying true to their unique personalities was a whole different problem. We had to carefully design the messaging logic so that each agent not only received messages but also responded in a dynamic way rather than feeling repetitive.

What we learned

We learned how to deeply integrate uAgents, Flask APIs, and LLMs to create responsive and entertaining AI ecosystems. This project sharpened our skills in orchestrating multi-agent conversations, managing async human inputs, and building a playful yet functional user interface. We also learned how crucial it is to lean into each team member’s strengths when building ambitious, multi-faceted projects.

What's next

We’re excited to evolve the agent personalities even further, we plan to add voice support (so you can literally talk to your agents), customizable team compositions (choose your dream startup team), and collaborative features so multiple users can co-pitch and watch agent dynamics unfold live. Long term, we imagine this tool becoming the ultimate toolbox for creators, founders, and dreamers everywhere.