- ChatGPT, Claude, Copilot, Gemini
- One connector for every assistant
ChatGPT, Claude, Microsoft Copilot and Gemini can now reach into other systems through the Model Context Protocol (MCP), an open standard all four support. One well-built connector lets your team ask any of them about orders, customers, policies or cases, without copying data into a chat window.
We build that connector, wire it into the assistants your organisation uses, and control what each person can see and do. We already run MCP servers for our own accounting and clinical costing software. For software companies we go further: we build your product as an app inside ChatGPT and Claude and submit it to their app directories, as we did for Saff and for Shog Consulting's corporate training platform.
What is included
- MCP server for your systems. One connector for your CRM, ERP, EMR, document store or database, usable in ChatGPT, Claude, Copilot Studio and Gemini Enterprise.
- ChatGPT apps and custom GPTs. Interactive apps built with OpenAI's Apps SDK, or custom GPTs with actions, for your staff or for your customers.
- Claude connectors and Skills. Custom connectors and Skills, so Claude works with your data and follows your own procedures.
- Copilot Studio agents. Agents inside Microsoft 365, Teams and SharePoint that use the same connector.
- Gemini Enterprise. Connections and Gems for organisations that work in Google Workspace.
- App directory submission. For software companies, we prepare your app's listing and submit it to the ChatGPT and Claude app directories. Quoted separately.
- Permissions and audit. Single sign-on, each person sees only what the source system allows, read-only by default, approval before any action that writes, and a log of every call.
Each assistant runs on its provider's infrastructure under your plan's terms. Where data must not leave your environment, the same connector can serve a private model instead.
How the engagement runs
- Scope. The quotation on this site captures your systems, the assistants your teams use and what they should be able to do.
- Build. We build the connector and test it against real questions from your team.
- Connect. We add it to ChatGPT, Claude, Copilot or Gemini with the right permissions.
- Submit. For products, we submit the app to the directories and see the review through.
Questions
What is an MCP server?
MCP, the Model Context Protocol, is an open standard for connecting AI assistants to other systems. An MCP server is the connector: it exposes your data and actions to the assistant, with your rules on who can use what.
Can one connector work in ChatGPT and Claude?
Yes. ChatGPT apps, Claude connectors, Microsoft Copilot Studio and Gemini Enterprise all support MCP, so one server can serve every assistant your organisation uses.
Does our data go to OpenAI or Anthropic?
When an assistant uses the connector, what it reads becomes part of that conversation and is processed under your plan with that provider. We return only what each question needs, and for data that must stay inside your organisation the same connector can serve a private model.
What is the difference between a custom GPT and a ChatGPT app?
A custom GPT is a configured ChatGPT with instructions, files and actions. A ChatGPT app, built with the Apps SDK, can show its own interface inside the chat and can be submitted to ChatGPT's app directory.
Can you get our product into ChatGPT and Claude?
Yes. We build the app, prepare the listing and submit it to the ChatGPT and Claude app directories, as we did for Saff and Shog Consulting's training platform. Submission is quoted separately, and each directory decides on approval.
Do you still build ChatGPT plugins?
OpenAI retired ChatGPT plugins. Their replacements are apps, connectors and custom GPTs, and those are what we build.