Getting started

Get started with DataTether

This page takes you from a new account to your first safe AI answer. You request access, open the console, connect a service, preview what gets generated, point an AI client at the hosted endpoint, and ask a question that reads real data behind your access rules. Each step shows exactly what you will see.

  1. 1

    Request access

    datatether.com -> Request guided access

    DataTether is currently in guided access. Start at datatether.com and choose Request guided access. Tell us about your organization and the OData services you want to connect, and our team provisions a workspace for you.

    What you'll seeAccess granted

    Once your request is approved:

    • Your organization gets its own console workspace.
    • You receive a sign-in link for that workspace.
    • Your access status changes to active, and you can open the console.
  2. 2

    Open the console

    Console

    Sign in to land on the DataTether console. This is the single place where your team registers services, configures what an AI client may see, and publishes hosted endpoints. Nothing is exposed until you choose to publish it.

    What you'll see

    The DataTether console after sign-in.

    This is where the product screen for this step will appear.

  3. 3

    Connect your first service

    Console -> Add Connection

    From the console, choose Add Connection to register your first OData service (for example SAP Gateway, SAP S/4HANA, or a public sample). DataTether reads the service metadata and generates a typed set of MCP tools, with read tools created first and write tools left off until you turn them on.

    This is just the short version. The full flow, including authentication, metadata discovery, and exposure settings, is covered in depth on the Connections page.

    What you'll seeTools generated

    After the service is registered you get:

    • A capability inventory of the service entities, keys, and fields.
    • Typed MCP tools generated for each entity you keep.
    • Read tools ready to use, with write tools off by default.
  4. 4

    Preview tools and UI

    Console -> Configure

    Before publishing, preview what you generated. You can inspect each tool, run it against the live service to confirm the output, and see a bound UI built on top of that output. This is where you decide which entities and fields are exposed and which operations are allowed.

    What you'll see

    Previewing generated tools and a bound UI before publishing.

    This is where the product screen for this step will appear.

  5. 5

    Connect an AI client

    Console -> Connect

    Publishing gives you a hosted MCP endpoint scoped to your organization and service. Point any MCP-compatible client at it over Streamable HTTP, with SSE compatibility for older clients. The endpoint exposes only the tools and resources you chose.

    Client-specific setup, including Claude, ChatGPT, and custom agents, is covered on the Code usage page.

    What you'll seeLive endpoint
    POST https://mcp.suriz.in/{org}/{system-name}/mcp

# Works with any MCP client over Streamable HTTP.
tools/list       list the tools you exposed
tools/call       run a tool with arguments
resources/list   list schemas and documentation
  6. 6

    Ask your first question

    Console -> Connect

    Ask your connected AI client a question in plain language. It selects a read-only tool, calls the hosted endpoint with the right arguments, and returns a structured answer your team can see. Every call runs behind your access rules, so the client can only read what you exposed.

    What you'll seeRead-only

    For a first question you get back:

    • A structured answer drawn from a read-only tool, not a guess.
    • A clear view of which tool ran and what it returned.
    • An answer scoped to the entities and fields you chose to expose.

Related