API v2 and MCP: A Powerful Combo That Opens the Bicom Platform to What Partners Build Next

A talk with Vesna Glavas Hrustic, Head of Product

Vesna Glavas Hrustic

Head of Product

No software vendor can build every solution for every customer, every industry, every workflow someone might dream up. That has always been true, and it is quickly becoming the fact partners most need to plan around. Bicom’s answer is a pairing: a native Model Context Protocol (MCP) server, live in v8.1 in gloCOM with PBXware support on the way, combined with a much stronger REST API, API v2.

Neither one tells the full story by itself. Together, they are the foundation for AI agents, custom applications, automations, and integrations that Bicom was never going to build alone.

Vesna Glavas

Let’s start with the big claim. Why is this bigger than any single capability you could point to on the platform?

Because a single capability is just that: one thing the platform can now do. This is a foundation. No software vendor, us included, can build every solution for every customer, every industry, every workflow a partner might dream up. 

We’ve always known that. What API v2 and the native MCP server change is who gets to build the rest. Right now, when a partner or their customer needs something specific that we haven’t built, that gap either doesn’t get filled or it becomes a ticket for us.

With this in place, that gap becomes something a partner’s own team, or their customer’s team, or an AI agent acting on their behalf, can close on their own. This, for sure, is not another feature but a completely different relationship with the platform.

coneptual photo, stairs arch

People hear API and MCP and assume they’re basically the same idea wearing different clothes. What’s the actual difference?

An API is built for developers. It gives you defined ways to request information or perform an action. You read the documentation, you pick the right endpoint, you write the code, and you decide what to do with whatever comes back. That’s been true of APIs for decades, and API v2 is a much better version of that same idea for the Bicom Platform.

MCP is built for AI agents, not developers writing code by hand. It gives an AI agent a standardized way to discover what tools and capabilities exist, understand how each one works, and pick the right one for whatever task it’s been asked to do. 

api vs api+mcp

The “USB-C for AI” comparison gets used constantly to describe MCP. Is that the best way to think about it?

It gets the most important part right: one standard connector instead of a different custom cable for every device. Before something like MCP existed, connecting an AI agent to a new system meant custom integration work for that specific agent and that specific system, every single time. 

MCP means an AI agent that already knows how to speak MCP can plug into any system that also speaks MCP, including ours, without custom glue code built just for that pairing.

Where the analogy runs out is that USB-C just moves electricity and data. MCP moves understanding, since it’s the format that lets an agent understand what’s on the other end of the wire well enough to use it correctly and safely, which is a much bigger job than a physical connector has to do.

I’d actually point to HTTP as the better analogy, both from the problem it solves and for how it works. HTTP standardized how browsers and servers talk to each other across the web. MCP does the same job for AI, it standardizes how agents discover, understand and interact with external systems and tools.

balance scale

Why do you need both API v2 and MCP? Why can’t one of them just do this job alone?

Because they’re solving different halves of the same problem. API v2 provides structured access to the actual capabilities and data inside the Bicom platform: the calls, the messages, the configuration, the events. That’s the functionality. 

MCP takes that functionality and makes it understandable and usable by an AI agent, describing what each capability does and when to reach for it. 

Put simply, the API provides the functionality, and MCP helps an agent understand what it can actually do with that functionality. 

Without API v2, MCP would have nothing real to point at. Without MCP, API v2 would still require a developer sitting between every AI agent and the platform, translating by hand. Together, they remove that middleman.

api mcp pbxware flow

Make this concrete. What could a partner actually build with this that they can’t build today?

  • A specialized agent for a specific industry: something that understands the particular call-handling and compliance needs of, say, a medical practice or a law firm, and acts on the platform accordingly, without a partner writing custom integration code for every single client that fits that profile. 
  • A connection between Bicom Systems products and a partner’s own internal systems, built and maintained by the partner.
  • A customer-facing application where the AI agent itself is the product, not just a feature bolted onto one. Automated operational processes that used to require a person watching a dashboard and reacting. 

 

Tools built to solve a need that’s specific to one partner’s market and would never have made it onto our own roadmap, because we can’t see every market the way each partner sees theirs.

 

podium ring

Native MCP support is shipping in gloCOM v8.1. PBXware support is coming. Walk through what changes when that lands.

gloCOM is where a user’s own communications data lives: their chats, their calls, their meetings. That’s the natural first place to prove that an AI agent can act usefully and safely against real data. 

PBXware is the system underneath, closer to configuration, system-wide call data, and administration. When MCP support reaches that layer, the same standard connector that lets an agent help one user manage their day extends to letting an agent help a partner manage their whole setup: provisioning, routing logic, reporting, the operational layer that today still requires a person in the admin GUI. 

That’s a much bigger surface, and it’s the one that turns this from a nice user-level convenience into a genuine platform capability.

 

headset

So what does “done” actually look like for this vision? What are you building toward?

A platform where a partner or a customer can describe a problem to an AI agent, and that agent can reach into the Bicom platform through MCP, find the right capability through API v2, and solve it, whether that’s answering a question, automating a process, or building a new workflow, without needing custom integration work from us or deep platform expertise from them. 

Partners should start thinking concretely about what they actually want an AI agent to do inside their own operation or a customer’s, because when PBXware-level MCP lands, those who’ve already got a real use case in mind will move a lot faster than the ones starting from a blank page.

Native MCP server support is available soon in gloCOM, with support for MCP in PBXware on the roadmap. API v2 is live in v8.1 as well, and expanding. Reach out to your Bicom Systems contact to explore your next steps.

cubes