What is MCP?

Seema Amble on why enterprise software vendors will fight agent middleware commoditization: being reduced to a data store is a decaying business, not a growing one.

“Nobody wants to just be put in a corner and said your job is to just store the this SQL format for expense reports and do nothing more and then we're going to use you only for that and then by the way we're we're piping you through to some other tool to analyze expense reports because that's a that's not a growing business that's a decaying business.”

Replit, used as an autonomous coding agent that spawns sub-agents (including one on OpenAI Codex) for complex tasks and collaborates with Claude via MCP without direct human orchestration.

“When Claude on Opus talks to Replet on sonnet if it's a big feature, Replet brings in a sub agent called the architect and the architect it turns out runs on codecs on OpenAI.”

Internal MCPs act as a company's knowledge base and holy grail, and the most common AI security mistake is accidentally bridging them to personal or external MCPs, creating uncontrolled data exposure.

“The connection of internal MCPs that are kind of like the holy grail of the of the company, the repository, their knowledge base connected, sometimes, most cases, by mistake with either personal MCPs.”

MCP integrations are not sufficient for production background agent deployments, forcing teams to build custom ad hoc integration layers instead.

“To actually get the integration right and the get the right experience often times we found that we had to go build our own ad hoc things.”

Nati Hazut on advanced teams replacing SIEM triage with AI-orchestrated incident response via MCP.

“The advanced customers, they're no longer consuming those alerts directly into their seam. Maybe they will do it for auditing, but how they're actually monitoring and doing the triage and the respond is through actually connecting our system with MCP along with other systems.”

In enterprise deployments, the MCPs actually connected to Claude Code were a superset of those formally approved, revealing an MCP sprawl problem analogous to shadow IT.

“Meaning that people were connecting cloud codes to custom MCPs. And you know, connecting it via like MCP servers to the ones that maybe weren't on that.”

Scaling context windows does not solve the tool-overload problem for agents because attention is not free, meaning smart tool selection remains essential regardless of window size.

“If you give 1,000 MCPs to LLM it's obvious like we are hitting like 1 million context windows. Maybe in some time it will increase to five or whatever, but attention is definitely not free.”

MCP (Model Context Protocol), originated by Anthropic, is emerging as a standard for representing services to AI agents and worth tracking as potential infrastructure for the agent economy.

“There's this acronym people are starting to use, MCP, that is a way for you to represent your service.”

Karan Vaidya on the cost of flooding an LLM with tools.

“If you give 1,000 MCPs to LLM it's obvious like we are hitting like 1 million context windows. Maybe in some time it will increase to five or whatever, but attention is definitely not free.”

Bold Security, an endpoint-based DLP and agent exfiltration detection vendor, is integrating via MCP to let AI orchestrate incident response, and claims a major resource efficiency advantage over legacy DLP tools.

“The advanced customers, they're no longer consuming those alerts directly into their seam. Maybe they will do it for auditing, but how they're actually monitoring and doing the triage and the respond is through actually connecting our system with MCP along with other systems.”