All projects

Zyro

10s to 3sproduct query time

Built Zyro's AI shopping assistant from scratch. A chat learns each user's taste and budget, then returns a few picks with a reason for each.

  1. React + TypeScript: the chat
  2. WebSockets: realtime replies
  3. Python services: taste, budget and picks
  4. OpenAI chat: the conversation
Fig. 1 Zyro went from proof of concept to working beta in 8 months, and product queries dropped from 10 s to 3 s.

The company

Zyro helps people make decisions online. It learns what each person likes “through conversation, not surveillance”. Then it gives “a small set of options, plus a quick explanation for each”.

What I built

I joined as the founding engineer and built the product from scratch. I built all of the user side: the React chat, the realtime WebSocket link and the Python services behind them. The chat learns each user’s taste, budget and priorities. It returns a short list of picks and explains each one.

The result

Zyro went from proof of concept to working beta in 8 months. Product queries dropped from 10 seconds to 3 seconds.

Options

Up, Down / W, S
Previous / next chapter
Left, Right / A, D
Select item
Esc
Close this menu