Stick

Inspiration

We built Stick out of pure frustration after realizing entrepreneurs were spending up to 80% of the week doing manual research instead of actually talking to customers. We needed to automate the soul-crushing part of prospecting so they could finally focus purely on closing deals.

What it does

Enter your company details and the products you're selling, and Stick's AI pipeline takes care of the rest. It scours the web, identifies your most relevant prospects, and for each match delivers in just minutes: a clear explanation of why that company fits your offer, a breakdown of which specific products they're most likely to buy, who is the most appropriate person to contact and their contact details, detailed analytics of the company as well as a cold call brief, a ready-to-send outreach email, and a fully generated consulting worthy pitch deck which is all tailored to that prospect.

How we built it

The whole system basically runs on a Python backend with FastAPI and the next year's frontend. We connect both of them via a WebSocket so that the Stripe API used to manage credits and the backend are properly connected. The whole agent is orchestrated by multiple cloud agents, each with access to multiple web search and web fetch tools provided by Linkup. We have used Paid.AI to fetch analytics on each agent's usage and how calculations are performed. And on top of that, we have a machine learning model that matches the ICP rubrics of the customer's current prospects and the prospects list that our AI agent fetches for them.

Challenges we ran into

We had multiple issues with the agent orchestration initially because we were making unnecessary, excessive calls to both Anthropic and Linkup. This was just making us run out of our credits that we actually paid for. We spent a couple of hours trying to find an efficient way to reduce these tool calls, and we achieved 90% savings, so it's great.

Accomplishments that we're proud of

Well, we're proud of a bunch of things, starting with our ability to finish the project with a multitude of features in such a short span of time, having worked with a team of four random people who had never met before. We had no idea till 3 p.m. yesterday, and we just started building after that. It was the first hackathon for more than half of the team, and everyone was just super excited and stayed awake. We shipped, shipped, and just shipped a gazillion features, and we have an end-to-end working product that, if we put it on a marketplace, I'm sure we'll find a couple of customers to use it right away.

What we learned

Well, I mean, what we learned is that it's not always that bad to team up with random people to work on random ideas. In the end, you always find some crazy shit. Each one of us learned something new:

The person who had experience with backend was trying to build some prompt engineering skills.
The one who had experience with the backend was trying to do something else. It was, in the end, more of a learning experience than anything else.

What's next for Stick

What's next for Stick? Well, time will tell about that, but hopefully we actually turn it into a great, great startup. We plan to stay in touch and actually make it bigger, like maybe Orange Slice or Clay, one of which has raised $5 million in seed funding, the other of which has raised money at a $3 billion valuation. I guess if we can build a product as well, not as good as them but almost as good as them or night, I'm sure we can do much more if we have more time.