Answering a simple question such as "why did checkout drop last week?" usually means opening three or four tools, each with its own login, its own filters and its own export. Henkan Chat puts those tools behind one conversation. This post describes what Chat does today, in beta, without the roadmap.
Connect your tools once
Chat works with the tools you connect, with your own permissions in each of them. The list covers the stacks we meet most often in e-commerce and experimentation: Google Analytics 4, Search Console, Google Tag Manager, Kameleoon, Wingify, AB Tasty, Contentsquare, Microsoft Clarity, Qualtrics, Henkan Vision, Slack and Notion, plus official vendor MCP servers and, on the Max plan, any remote MCP server you add by URL.
Credentials are tested before they are saved and stay on the server, encrypted; they never reach the browser. You can connect several accounts of the same tool, for example two Google accounts or a second Kameleoon workspace, and Chat names each connection so it never mixes them up. Connections belong to your organisation, so your team joins the same workspace without setting anything up again (personal connections such as Slack stay yours).
Answers you can check
You type a question in plain language. Chat picks the tools it needs, shows every call it makes, and answers with tables, charts and sources. Every table can be downloaded as CSV, and the charts are drawn from the same numbers as the table, so nothing in the answer comes from nowhere.
A/B test questions follow stricter rules. Chat reports the uplift, the 95% confidence interval, the p-value and a sample ratio mismatch check, and it only says a variation wins when the result is significant at 95%. Otherwise the answer reads "not significant yet" or "inconclusive (SRM)".
Some tools can change what your customers see: pausing an experiment, publishing a tag, posting to Slack. Chat never runs those without your approval. Each approval card says in plain words what is about to happen, with the state before and after, and whether the change is a draft or live. Live changes ask every time.
Flows for the work you repeat
A Flow is an investigation you would otherwise run by hand. You drag connectors onto a canvas, add conditions and branches, let a Henkan agent analyse what came back, then send a report by email, Slack or Notion. A Flow runs on a schedule, when an event happens in one of your tools, or when you start it.
Templates give a starting point. The conversion diagnosis template, for example, checks GA4 conversion every Monday and, when it drops, looks at the pages losing the most in Contentsquare and the tests started or stopped in Kameleoon before ranking the probable causes. Flows only read data: the only blocks that write are the Slack and Notion outputs you add yourself.
Wherever you already work
Chat is not only a web page. Mention @Henkan in a Slack channel or send it a direct message, and it answers in the thread with the same tools, statistics and approvals. Chat is also an MCP server: add it once to Claude, ChatGPT or Cursor, and that assistant can use all the connections you set up in Chat, with the same accounts and permissions, plus your Flows.
Chat is available to the teams of Henkan's clients, on a plan chosen with Henkan, and every organisation works with a Henkan consultant. If you already have access, the documentation covers connectors, approvals and Flows step by step. If not, talk to Henkan.
