BizzE

Inspiration

BizzE started when one of us tried to buy a small business. As a first-time buyer, the process was daunting. Due diligence felt like a wall: messy financials, unfamiliar terminology, and no clear way to know if the numbers actually held up. It was enough to make anyone walk away. That experience revealed a bigger problem. Serious diligence has always been reserved for private equity firms and institutional buyers with teams of analysts ready to tear apart a deal. The everyday buyer gets none of that. They either pay thousands for professional reports or trust their gut and hope. We built BizzE to change that. We wanted to give the individual buyer a fighting chance: real, evidence-backed due diligence where every claim traces to a source document or calculation, not a black-box verdict. Buying a small business shouldn't be a privilege reserved for institutions.

What it does

BizzE allows for the discovery of small businesses, and allows users to upload financial documents a SBO can provide on their SMB which get a full professional-level audit and breakdown in plain English to help the buyer/seller understand the risks and strengths of buying/selling a business.

How we built it

We started building this at a local hackathon in Dallas when we realized there was growth in this market. The idea of the hackathon was to use AI to develop an MVP for an idea in under a month - a very short time to develop a tool as sophisticated as this one. Using Codex and our loosely cobbled together iteration planning to build the platform piece by piece, we ended up with a heavily AI reliant first MVP built using FastAPI, Next.js, ts, React, Supabase, OpenRouter, Mistral, and Railway. Much of this infrastructure still lives on today, but we've continued to iterate in order to reduce the dependence on AI to do financial due diligence and to increase the ability to trust and scale the audit performance.

Challenges we ran into

Professionals in this space care deeply about how much they can trust the numbers. Given that AI is probabilistic, trust starts to slip when you ask AI to do a ton of calculations of its own accord and move that away from a static calculator's hands. We realized that the calculations in financial audits remain static, the main job of the LLMs we used were to extract the data from source documents so that they could be used in calculations. This is a challenge we ran into where we needed to build repeatable, reproduce-able outputs on audits. Another one we faced was with

Accomplishments that we're proud of

We're proud of finishing in second place in the hackathon we competed in a few months ago against many accomplished professionals in software development. We built the first hypothesis in a month, and we continued to garner interest from peers as we continued to build it. We're proud of building a solution which can cheaply and efficiently do what many people find a huge monetary loss in.

What we learned

Building the actual tool was a small part of the task, the bigger deal was figuring out exactly where it was that auditing professionals lose time on in their flows. We learned that the conversations we have with folks in this field is where we're going to get the most value out of creating a tool that could genuinely be useful in day-to-day workflows.

What's next for BizzE

We want to turn BizzE into a full-fledged business and source document discovery platform as well, so it would make finding and compiling necessary information to make a complete, well-informed deal on purchasing a business an incredibly simple and straightforward task that anyone can use. And as always, we want to improve the quality of our product as we build trust in the consumers.

Built With

• fastapi
• next.js
• openrouter
• python
• typescript