Here is an alternative and very convenient way to build a working voice or chat agent: open the tool you already write code in, describe what you want in plain language, and a few prompts later, you are talking to a live agent you can interrupt and try to break.
That is the OCP MCP Server. It lets you build, deploy, and talk to an Omilia agent from a chat prompt, inside the tools developers already use: Claude Desktop, Claude Code, Cursor, VS Code, Codex.
No clicking through a console. You describe the behaviour you want in plain language, and the assistant builds it. Bring an idea to a talking prototype in minutes, from your laptop or your phone, virtually anywhere in the world.
The reason it feels convenient is that the work happens in plain language. There is no separate interface to learn, no sequence of console screens to memorise. You say what the agent should do, the assistant translates that into the right OCP tool calls, and you test the result by simply talking to it. The loop from idea to feedback is a conversation, not a project.
Two things change when building an agent is as simple as having a conversation:
- Developers stay in their workflow. OCP work happens without leaving the environment they already write code in, alongside the rest of the build. No context switch, no second tool to keep open.
- The builder pool widens. You no longer need a platform specialist to build a working agent. Anyone who can describe the behaviour they want can produce one, from product managers to solution consultants to engineers who had never opened the console.
The business payoff is speed where it counts. Demos and proofs of concept reach customers far faster, because the gap between an idea and a live agent has collapsed. When buyers want to see something real rather than a deck, time to first prototype is a competitive edge. It also arrives as MCP itself becomes the common standard for connecting AI to tools, now backed across OpenAI, Google, Microsoft, and AWS, so OCP meets developers where the wider ecosystem is already heading.
One thing has not changed, and should not. The LLM in your MCP client drives the build experience, not the production ready agent. Deployed agents run on OCP’s auditable, SLM based architecture. Omilia uses LLMs offline to learn from real conversations, and keeps them off the live call. That distinction matters most to regulated buyers, who need to know that nothing is improvising responses while a customer is on the line. The build got conversational. The runtime stays disciplined.
Building agents should be fast, and open to more than a handful of specialists. Now it is both. Your next demo can start as a sentence in a chat window and end as a real agent a customer can talk to. For more information check out our changelog on the new capabilities.
About the Author
Conall Murtagh, Product Marketing Manager
Conall is a trusted product marketing professional with over 20 years of experience in B2B technology. With a cross-industry track record of shaping how complex products are understood in the market, he has a genuine interest in how technology can help the world communicate better. Based outside Madrid, he is a keen amateur chef and enjoys running, cycling and spending time with his wonderful family.
