BizToken

Inspiration

Going to Stanford, we see all the investor hype around tech but we also see that small businesses are growing faster than ever, and many require capital to expand. We talked to small businesses and realized they often struggle to raise money (and investors are losing out on growing companies). Especially because these small businesses are central to our communities—people want to help and our website helps them do that in a market-driven approach.

What it does

We built a tool that connects small businesses with investors—firms, nonprofits, and accredited individuals—who want to invest in their community's "mom and pop shops" and make money while they're at at. Our platform has a few key features: 1) When small business entrepreneurs submit a request for funding that includes comprehensive questions about them—and a chatbot to make sure we're asking the questions that matter most to investors—we pass that request through our investor-like LLM agent that creates an analyst report and scores potential investments based on growth potential and risk. 2) In the investor portal, potential investors can find promising small businesses that want funding. 3) We create custom NFTs that act as fictional "fractional shares" of small businesses. The businesses are sent a link to control these NFTs so they can control who gets sent the shares (therefore controlling who gets stake in their company). 4) In the backend, we've incorporated a tool where investors can search for things that they want in an investment using augmented vector retrieval (example: "find me a burger shop in Palo Alto that's growing fast" will match them with just that if one exists in the data). In MVP, we have this almost completed fully.

How we built it

1) Front-End: we used Reflex to build our front-end. We built 3 pages: chatbot page (where small business entrepreneurs can interact with our investor agents), business page (where businesses can fill out a form for initial funding), and investor page (where investors can search for good matches for them). 2) Generated of synthetic training dataset using OpenAI LLMs (simulating economics models in this process) 3) OpenAI LLM for investor analysis report and scoring 4) Pinecone for vector db and search 5) Crossmint API for creating collections and tokens unique to each company

Challenges we ran into

First and foremost, none of us had much experience with front-end (which is where our project hit a roadblock). We had to figure out how to use Reflex to integrate with backend. Some Reflex features didn't function as expected, so navigating around that was difficult.

How we structured the project is that each team member generally focused a sector of the project, and linking these sectors together (due to Reflex's compatibility with all of our systems) proved to be difficult. The team's support was helpful but we remained challenged by this.

Crossmint was initially to get used to the Crossmint platform because we wanted a feature that they didn't directly support, but then the Crossmint team was very helpful in adding that feature for us.

Accomplishments that we're proud of

We think our product solves a legitimate market need. We're proud of our progress during the 2 day period, and we're excited to continue pursuing the project as a startup after the Hackathon ends. We're also proud of the fact that we were able to create each individual component of our project and work very nicely together to problem-solve and build something cool together.

What we learned

We learned a lot about the front-end aspect of development, using LLMs to our benefit, and about the use-cases of blockchain.

What's next for BizToken

We want to continue with this idea and work on it as a startup. We want to use this idea as a means to grow our knowledge about AI, LLMs, NFT, and investing, and we want to talk to more customers about how we can make it better for them.