Flippit: AI-Powered Reselling Automation
Inspiration
Back in highschool, we used to be big into reselling shoes and watches. Unfortunately, it was WAY too much of a timesink.
If you think about it, the entire process is very cumbersome and complicated:
- 4-6 hours sourcing and compiling vendor lists
- 3-5 hours calling and negotiating each and every single one of them
- 2-3 hours deciding on final prices without market context
- 1-2 hours creating descriptions for listings and posting on secondary marketplaces
- 10-16 hours total - I'm getting exhausted just thinking about it
We were spending entire weekends on this stuff. That's when we realized: what if AI agents could handle this entire workflow while we sleep? So, we built Flippit.
What it does
Flippit is a symphony of dozens of agents working in tandem that transforms reselling from a 12-hour manual grind into a 12-second automated process.
First, the user tells Flippit the product he/she wants to flip a profit on. Flippit deploys an agent to compile a list of the most relevant product vendors in the city, their location, and their phone number.
Then, Flippit deploys voice agents to call each and every one of these vendors sequentially. We're able to inform each agent about the results of the earlier calls, meaning each agent is able to negotiate with the context of the previously lowest offered price + bring the price down even more.
Flippit uses one last agent to compile all the call info, calculate a reasonable sell price, and create listings on secondary marketplaces such as EBay, Craigslist, and Facebook Marketplace.
Total time saved per flip: 15+ hours
How we built it
The agent that compiles a list of relevant vendors is powered by Perplexity. We return an object from these LLM calls that we then upload to our supabase for storage.
The agents that autonomously call each vendor are powered by vapi.
Lastly, the agent that compiles the info and creates listings is powered by GPT/EBay's listing API.
Challenges we ran into
Figuring out a way to get our voice agents to carry out context-based negotiation was a hassle.
Accomplishments that we're proud of
We are proud of the immense value we're able to create by automating the reseller's process. By building an agentic workflow, we were successfully able to entirely automate a very tedious process and deliver a solution which has the potential to dramatically alter the entire reselling industry.
The vendors get to sell more of their product, the buyers get to enjoy more of the type of good they like, and the resellers benefit because they're able to bring in a higher volume of deals/more revenue into their business.
What we learned
This was one of our members' first hackathons, and he ended up contributing a ton, so we were really glad to have him aboard.
What's next for Flippit
We intend to ask our dropshipper friends to see if they would use the product. If there's certifiable demand, we'd then get started on building and deploying a more robust MVP and look to scale Flippit into a full-fledged company.
After dominating the resell industry, we'd look to take over the manufacturing and customer sectors next.
Built With
- express.js
- gpt
- nextjs
- perplexity
- supabase
- tailwind
- vapi
Log in or sign up for Devpost to join the conversation.