Claude
- In Claude, open Settings → Connectors → Add custom connector.
- Paste the server link.
https://mcp.absee.it/mcp - Sign in to ABSee when asked. That's it.
ABSee / MCP
ConnectAdd https://mcp.absee.it/mcp as a connector in Claude, ChatGPT or Gemini and sign in to ABSee. It takes about a minute; then ask about real experiments in your own words.
Updated October 2026 · Source: ABSee record · Methodology →
ABSee is experimentation intelligence for ecommerce: a record of real experiments run by 1,150+ brands, with the page before, the version tested, and whether it was kept. It plugs into your assistant over MCP, the open standard assistants use to call outside tools. There is no dashboard to learn and no integration project.
https://mcp.absee.it/mcpclaude mcp add --transport http absee https://mcp.absee.it/mcp{
"mcpServers": {
"absee": { "url": "https://mcp.absee.it/mcp" }
}
}Menu names differ between assistants and change often; look for “connectors”, “tools” or “MCP”.
You need an ABSee sign-in; you can start at app.absee.it/start.
The first time you connect, your assistant asks whether you are a brand team, an agency or a consultant, and which site you work on. That is how the answers come to be about brands like yours. The last two jobs, and the outcome on each experiment, are part of the Intelligence plan; on other plans your assistant answers with a search instead.
What changed in fashion and footwear this week?
Comes back with experiments matched by idea, page type, device, brand, category or date, each with a before-and-after thumbnail.
Show me that one full size: the page before and the version tested.
Comes back with the full-size page before and version tested, what changed, and related experiments.
Which of my competitors does ABSee have experiments for?
Comes back with which of the brands or domains you named ABSee has experiments for.
We sell premium outdoor footwear online. Which brands should we compare ourselves with?
Comes back with close competitors that ABSee has plenty of experiments on.
Our hypothesis: if we collapse the content around the buy button, then more shoppers will add to bag, because there is less to distract them. Who tried it, who kept it and who rolled it back?
Comes back with comparable experiments other brands ran, filed as evidence for, against, or closest match.
Look up the free-shipping progress bar pattern. Where do brands put it?
Comes back with what the pattern is, where brands placed it, what they did next, example experiments and the principle behind it.
Each result carries the brand, the page type, the dates it ran, and before-and-after images your assistant can show you. Your assistant is told never to state an outcome or a statistic that ABSee did not return, so if nobody has tested an idea, that is the answer.
Example answer, shortened.
Yes — Timberland tested it, and kept it. It added a line to its mini-cart warning that items in the bag are not reserved, seen from 19 May to 4 October 2026, and the tested version became the page.
The takeaway on the record: stating that carted items are not reserved plays on loss aversion, prompting faster checkout decisions. Here it is:
Before
TestedWhat changed Adds a line warning that items in the bag are not reserved.
Seen 19 May to 4 Oct 2026
ABSee does not see your conversations with your assistant, only what the assistant sends when it calls an ABSee tool. Your own data stays private to your team and is never shared or added to the market evidence. The full detail is in the privacy policy.
https://mcp.absee.it/mcp. Add it as a custom connector in your assistant, then sign in to ABSee.
Claude, ChatGPT or Gemini, and any assistant that supports MCP connectors. Claude Code connects with one terminal command.
The Model Context Protocol, an open standard that lets an AI assistant call outside tools. ABSee uses it so your assistant can search real experiments while it answers you.
Yes. The first time your assistant uses ABSee it opens the ABSee sign-in. You can start at app.absee.it/start.
No. ABSee only sees what your assistant sends to it when it calls an ABSee tool, such as the question it is searching for.
What's been tested against fit uncertainty on mobile product pages?
Works in Claude, ChatGPT or Gemini once ABSee is connected.
Connect your AI →