Placeholder

Github repo: https://github.com/kosuvorov/synthetic-shopping-agent-mvp

Inspiration

Online shopping is fragmented and time-consuming. We wanted to turn it into one calm conversation: describe what you need, confirm your preferences, and let an agent handle the search.

What it does

Placeholder searches for products, compares prices, specifications, delivery, returns, and store trust, then recommends the best option. Users can buy at the current price or monitor it and automatically purchase when it reaches their target price. Voice, typing, and clickable options remain synchronized throughout the experience.

How we built it

We built a responsive web app with an AI-guided shopping flow, OpenAI Realtime voice interaction, structured product analysis, live product discovery, and Firecrawl-powered price monitoring. The interface presents complex shopping decisions as concise, accessible steps.

Challenges we ran into

The hardest parts were grounding AI responses in the displayed product data, handling incomplete or ambiguous requests, finding reliable product images and pricing, and keeping voice interactions synchronized with visual controls without overwhelming the user.

Accomplishments that we're proud of

We created a fully working, deployed agentic commerce experience with multimodal interaction, product validation, comparison, price monitoring, auto-buy controls, and support for both physical and digital products—all within a minimal, mobile-first interface.

What we learned

Trust matters more than raw automation. A shopping agent must explain what it found, avoid inventing details, ask only necessary questions, and give users clear control over purchasing decisions.

What's next for Placeholder

Next, we'll expand retailer coverage, improve real-time price verification, add secure checkout integrations and notifications, and launch Placeholder on Telegram, the App Store, and ChatGPT.

Built With

• a
• chatgpt
• claude
• codex
• great
• team