VentureVision

Validate before you build

Comment 1

Inspiration

Every week a friend had “the next big thing,” a hoodie swap, a smart water bottle, compostable confetti, and more. Yet no one had data to know if it was worth a weekend, let alone life savings. After watching three ideas crash and burn (and more than 1000 dollars disappear), I joked that we needed an “AI co-founder” to call out the delusions before the first dollar was spent. The joke stuck, and VentureVision was born.

What it does

  1. You input your business idea.
  2. Returns a market score (0–100 %), Year 1, Year 2, and Year 3 revenue bands, a revenue forecast, and a detailed list of reasons why you got your success score.
  3. Provides a ranked recommendations list that can further boost your success rate
  4. The chatbot answers “how do I price this?”, “who is the real customer?”, or your custom questions.
  5. Lastly, it deep searches the web to find your future business competitors.

How we built it

I began by writing a simple, step-by-step plan in a Base 44 doc, which models to try, and how the pages should flow. Then pasted into Base44, where it created the rough draft of this model. After, I pasted the starting code onto github and over the span of weeks, tweaked it to fix the issues and meet my requirements.

Challenges we ran into

  1. The rough draft was positively optimistic, creating unrealistic success rates. I had to train the forecasting engine to be brutally honest by inputting examples and creating several guidelines that the AI engine would follow in order to calculate the market score.
  2. The model then wasn’t able to correctly differentiate the difference between a weak, moderate, and good business idea, so I had to further adjust the algorithm to correctly reflect the true potential of the business plan.
  3. The forecasting revenue was also a problem as it was unrealistic. I had to then adjust the revenue template to directly correspond to the success rate.
  4. When testing, the “mystery stain hoodie and QR code to previous owner’s browsing history” idea scored 55% before I inserted a legal risk guideline that dropped the score to <25% due to health and privacy hazards.
  5. Finally, there were many front-end and design problems.

Accomplishments that we're proud of

  • The chatbot feature
  • The simple and futuristic design
  • The webscan feature

What we learned

  1. In the process, I was able to refine my python skills and learn how to efficiently work with AI.
  2. While editing my demo video, I learned many new skills that I will definitely use in the future.

What's next for VentureVision

I am currently creating a…

  1. Competitor API, when plugged into shopify and amazon, users can see live price gaps
  2. Voice to idea feature, mobile where users can record or import live pitches
Share this project:

Updates