Production web agents have a dirty secret. Most of their token budget goes to parsing bloated HTML — scripts, tracking pixels, ad divs — before the model sees the actual content. Web Speed, launched on Product Hunt on May 9 by Dominic Pi-Sunyer, attacks this head-on: a logic layer between the agent and the web that translates pages into token-efficient JSON maps. Claimed savings: 70-90% per action.
What it actually does
A compiler for the web. Instead of feeding Claude or Gemini a 50KB HTML dump, Web Speed pre-digests the page into structured JSON the agent can navigate with a fraction of the tokens. Less input, faster decisions, smaller bill. The product is built for agents, not humans — there is no front-end to click around. It’s plumbing for the action layer.
SDK and MCP server
Two entry points. The SDK drops into any custom agent stack. The MCP server plugs directly into Claude, Gemini, crewAI, and anything else speaking MCP — your agent gets web access without wiring up Playwright or a custom scraper. Typical uses: form filling, price monitoring, multi-step research, checkout flows, any task where the agent needs to actually move through a site rather than read a single page.
Why now
Six months ago, “AI agent” meant a chatbot wrapper. Now buyers want agents that take actions, and cost per action is the limiting factor. Compute keeps getting cheaper; the API bill of an agent re-reading the same product page ten times does not. Token efficiency at the web layer is the new bottleneck, and Web Speed is one of the first to name it explicitly.
You Might Also Like
- Agent Action Protocol aap the Missing Layer Above mcp That Actually Makes Agents Production Ready
- Jupid hit 1 on Product Hunt During tax Season 507 Upvotes for an ai Accountant Built on Claude Code
- Insforge Hits 1 on Product Hunt and 3600 Github Stars is This What Agent Native Backends Look Like
- Mulerun hit 1 on Product Hunt but can an ai Agent Marketplace Actually Work
- Design Agent by Lokuma Topped Product Hunt With a Simple Pitch ai Agents Need Designers too
Top AI Product 
Leave a comment