What is an MCP server?

MCP (Model Context Protocol) is the open standard AI assistants use to connect to outside products. An MCP server exposes your product's actions as tools the assistant can call, with the user's permission. ChatGPT apps and Claude connectors are both built on it, so one server can serve both.

Is a ChatGPT plugin the same as a ChatGPT app?

Yes. OpenAI calls them apps, built with the Apps SDK, and ChatGPT now lists them under Plugins. Either way, it is an MCP server plus a listing that has passed OpenAI's review.

Can one connector work in both ChatGPT and Claude?

Yes. The same MCP server works in both, and in other MCP clients such as Cursor and VS Code. Each directory has its own listing and review, which we handle for you.

Can the connector stay private to our team?

Yes. A connector does not have to be listed publicly. Your workspace admin can add it for your team only, and it can be limited to reading data, with an audit log of every request.

How long does it take?

Most projects roll out in 3 to 4 weeks from the start, depending on how much of your API the assistant needs. A public listing then waits on the directory's review, which OpenAI and Anthropic control.

Who owns what you build?

You do, from day one: no lock-in, no hostage repositories. You get a clean repo with tests and docs, so whoever you hire next inherits an asset, not a liability.

ChatGPT apps, Claude connectors and MCP servers

Get your app into ChatGPT and Claude

We turn your product or API into a ChatGPT app (what ChatGPT now lists as a plugin), a Claude connector and an MCP server that works in any AI assistant, and we take it through each directory's review. We have done it for our own products 27 times.

Get it built

Developer typing beside a tablet of code and a hand-drawn flowchart of an integration

What we build

ChatGPT apps and plugins
Your product inside ChatGPT, built on OpenAI's Apps SDK and the Model Context Protocol, and listed in the ChatGPT plugin directory.
Claude connectors
The same product in Claude's connector directory, so it works in claude.ai, Claude Desktop, the mobile apps and Claude Code.
MCP servers for your API
A remote MCP server in front of your existing API, with OAuth sign-in and scoped tools, usable from ChatGPT, Claude, Cursor, VS Code and other MCP clients.
Private connectors for your team
Connect an internal system, such as a call centre, CRM or database, to your company's ChatGPT or Claude workspace. Read-only by default, with an audit log.
Directory submission and review
Listing copy, test accounts, tool annotations and the back-and-forth with OpenAI and Anthropic reviewers until the listing is live.
Hosting and upkeep
We host and maintain the server for you in Canada, or set it up in your own cloud account, and keep it working as the platforms change their rules.

Why put your product inside ChatGPT and Claude

More and more people start their work in an AI assistant instead of a browser tab. A listing in the ChatGPT or Claude directory puts your product where they already are: they connect their own account once, then ask the assistant to read, create or update things in your product in plain language.

For an internal team, the same technology does a different job. A private connector lets staff ask questions about your own systems, prepare reports and spot exceptions without exporting data into spreadsheets first.

Proof: our own products are in both directories

Every listing below is one of our own products, built, submitted and maintained by our team. That is 14 plugins in ChatGPT and 13 connectors in Claude, each through the platform's own review.

Many agencies now list MCP as a service. Before you hire one, ask to see a server running in production. Ours are public, so you can open any of them and see the result.

How a connector project runs

You describe your product or system at /start. An engineer replies within one business day with a scoped plan. The main scope question is which parts of your API the assistant should see, because each part becomes a set of tools to design, secure and test.

From there we build the server, wire up sign-in, test it in the real ChatGPT and Claude clients, and prepare the listing. Most projects roll out in 3 to 4 weeks from the start, depending on scope. Directory review time is set by OpenAI and Anthropic, so we submit as early as the build allows.

Two developers at a desk reviewing code on a laptop, one pointing at the screen

What makes a connector pass review and stay useful

Reviewers check that every tool has a clear name and description, says whether it only reads or can change data, and asks for confirmation before anything destructive. They also need a privacy policy, a support contact and a working test account. We set all of this up from the first build, not after a rejection.

After launch, the work is keeping the connector useful: least-privilege scopes, sensible limits, logging that shows what the assistant did, and updates when OpenAI or Anthropic change their requirements.

Our listings in ChatGPT and Claude

Open any of them to see a connector we built and maintain. Each opens in the official directory.

SuperBooks plugin page in the ChatGPT plugin directory, with an Install plugin button and three example prompts — SuperBooks in the ChatGPT plugin directory

SuperBooks connector page in the Claude directory, made by Devino Solutions, with its connector URL — The same product in Claude's connector directory, made by Devino Solutions

ChatGPT: 14 plugins

Claude: 13 connectors

Case studies

GetItDone — a task manager your AI can actually see

An AI-native task manager where daily plans, stand-ups, tasks and time tracking live in one place, and AI agents get the same task context a person does.

SuperBooks — AI bookkeeping for small businesses

AI bookkeeping and accounting software that connects to your bank, categorizes each transaction and matches it to its receipt or invoice.

Postify — one post, tailored for each social platform

A social media scheduler: write a post once, tailor it per platform with an AI copilot, then schedule or publish it to your connected channels.

Frequently asked questions

Prices are fixed and shown up front: pick what you need on /pricing and the total and timeline update with every answer. There is no hourly billing, no setup fee and no project-management fee. For anything the configurator does not cover, send a brief at /start and the reply includes a scoped plan within one business day. For market ranges and what drives cost, read the guide at /software-development-cost-canada.

More services

Custom Software Development SaaS Development Web & Mobile Apps Software Consulting AI Development

AI development AI development services Our apps How to choose an AI development company

What are you building?

Three minutes at /start: describe the project. An engineer replies with a scoped plan within one business day, not a sales call.