It’s been a busy week in MCP land. Last week brought two big announcements in the MCP ecosystem:
- The launch of the official MCP Registry in preview.
- OpenAI’s announcement that MCP support has been added to ChatGPT.
I invited Craig Jellick to join The Context this week and chat about these two developments and what they mean for developers, IT leaders, and enterprises looking to adopt AI responsibly.
Why the MCP Registry Matters
The new MCP Registry provides a global clearinghouse where anyone can publish an MCP server. In just a few days, over 160 servers have already been registered. That’s a good sign of momentum.
The idea is simple: the upstream registry is open and uncurated, ensuring maximum participation. Anyone can push an MCP, and minimal validation (like DNS or repo ownership) is used to indicate authenticity. Downstream, organizations, vendors, and platforms can build their own curated registries — trusted catalogs where quality, security, and compliance are enforced.
This registry unlocks the possibility of a standardized discovery layer for MCPs, but as the video discussion notes, adoption depends on clients and downstream registries building on top of it.
Where Obot MCP Gateway Fits In
The registry is an important foundation — and will help organiations as they build out their own private MCP catalogs. Ideally enterprises will be able to identify authentic MCPs developed by organizations they trust, and incorporate them into a private registry that also imeets other key needs, such as:
- A way to curate MCPs for their organization.
- Access control tied into identity platforms like Okta, Entra, or Google.
- Visibility and governance across usage, performance, and compliance.
- A user-facing catalog that employees can actually browse and connect from.
That’s where the Obot MCP Gateway comes in.
Obot acts as a downstream registry and control plane for enterprises. IT can take MCPs from the global registry (or from internal sources), onboard them into Obot, and apply governance policies. Users then get a trusted directory of MCPs, complete with documentation and secure connection URLs for their clients and agents.
And because Obot proxies all traffic, IT maintains control and visibility while employees gain frictionless access to the MCPs they need.
Why ChatGPT Support for MCP is a Big Moment for Enterprise AI
With MCP now embedded into ChatGPT, we are primed to see even more growth in MCP server development and even agents that are exposed as MCP servers. If your organization is using Enterprise ChatGPT, this is the ideal time to begin looking at an Enterprise MCP Gateway. It’s a big opportunity to encourage adoption of this new technology, and avoid pitfalls of uncontrolled MCP consumption, such as:
- Shadow MCPs proliferating with no oversight.
- Insecure MCP adoption from unverified servers.
- Adoption bottlenecks if users can’t easily discover and connect MCPs.
Obot solves this by bridging open innovation with enterprise control. We’re excited to see the registry launch, and even more excited about helping organizations safely embrace the rapid growth of MCP adoption.
Watch the full discussion here:
https://youtu.be/YCLgeNUyMuE?si=RFe-NYCZUKb5UIWG
