Skip to content

VEDUTUM in your AI assistant

Ask Claude or ChatGPT how many points a set of renderings needs, where the shadow falls on the site on 21 December or whether the export from your CAD program will work — without a VEDUTUM account. The assistant fetches the answer from us through an open interface (MCP).

What the interface does not do: it makes no images or videos, creates no projects and has no access to your account. All of that stays in the software.

Seven tools

Every tool only reads. None calls an AI model, costs points or changes anything. The assistant picks the right one; you simply ask.

Links in the answers carry a tag for the assistant (such as utm_source=claude and ref=ki-assistent), so that we can count how many sign-ups come this way.

Claude

Works now on every plan, including the free one, which allows one custom connector. The quickest way is the link, which opens the dialogue with name and address already filled in:

Add VEDUTUM to Claude

Or by hand:

  1. In Claude, open Settings → Connectors.
  2. Choose “Add custom connector”.
  3. Name: VEDUTUM, URL: the address below. No sign-in is needed.
  4. Add. In a chat, the connector can be switched on and off in the tools menu.
https://www.vedutum.com/mcp

In Team and Enterprise organisations, the owner of the organisation adds the connector for everyone; members then switch it on for themselves.

ChatGPT

VEDUTUM has applied to ChatGPT’s plugin directory. Until it is listed, users on Pro, Business or Enterprise can add the server as an app in developer mode, using the same address:

  1. Turn on Settings → Apps (“Plugins”, depending on the version) → Advanced → Developer mode. In Business and Enterprise, the administrators decide whether developer mode is available.
  2. Choose “Create app”, name it VEDUTUM and enter the address above as the URL.
  3. Authentication: none. Save, then select it in a chat.

Whether and when plugins are available for accounts in a given country is up to OpenAI’s rollout; we have no say in it.

Gemini

Today Gemini accepts its users’ own servers only in the United States (Gemini Spark) and in Gemini Enterprise, through a company’s administrators. Anyone with access there uses the same address; for everyone else, Gemini is not yet a route to VEDUTUM.

Other MCP clients

Cursor, VS Code, Claude Code and any other client that speaks MCP over Streamable HTTP connect with the same details: the address above, transport Streamable HTTP, no authentication. Claude Desktop uses the connectors from your Claude account, so the steps above are enough there. For clients with a configuration file:

{
  "mcpServers": {
    "vedutum": {
      "type": "http",
      "url": "https://www.vedutum.com/mcp"
    }
  }
}

VS Code expects the same entry under servers instead of mcpServers.

Privacy

  • What reaches us: only the details the assistant sends for a tool — a town or postcode for the sun position, the number of images, a file type. For the sun position, town and postcode are enough; an exact address is not needed.
  • What we do not store: neither the details nor the answers, no cookie. The IP address of the calling server — usually the platform’s server, not your connection — is held only in the server’s memory to allow at most 120 calls a minute. The program identifier (user agent) tags the links by platform; all we count, per day, is which tool was called from which platform.
  • Place lookup: for the sun position, our server sends the place to OpenStreetMap’s Nominatim service; the service sees our server, not you.
  • The conversation itself is between you and the platform, and its privacy policy applies. Ours is in the privacy notice (German), section 6.

Contact

Questions about the interface: office@vedutum.com. How VEDUTUM uses AI in its own product is set out on AI transparency.

Start for free