Launched this week
Periskop
Product discovery MCP for AI agents, across the open web
8 followers
Product discovery MCP for AI agents, across the open web
8 followers
AI agents can chat, but they're bad at shopping: they scrape pages, burn tokens, and still return links you can't trust. Periskop is the product-discovery layer for AI agents: one MCP/API call turns a shopping intent into ranked products, alternatives, and bundles from the open merchant web, with prices, images, and working merchant links. It stops at discovery: no checkout, no payments. Build shopping assistants, browser agents, procurement, gifting, and price monitors. Live now with API keys.
How does Periskop actually pull in working merchant links, and is there any safeguard when a retailer blocks scraping or changes its page structure overnight?
How does the ranking actually work under the hood, and do merchants pay to be surfaced or is it purely based on relevance and price signals?
How are you handling the freshness of prices and stock when scraping from open merchant pages, and is there a recommended fallback if a merchant blocks the request?
How do you handle merchants that block scraping or have inconsistent product feeds, and does that affect the reliability of the links your API returns?
Really clean idea, especially the part about not touching checkout. Tried it for a gift idea and the alternatives list was way more useful than what I usually get from a regular search.