what is MCP in plain English?

It stands for Model Context Protocol, and it is a standard plug. Without it an AI can only talk. With it, it can look things up in your software and, if you allow it, change them. The plug for one specific program is called a connector, and inside the connector is a list of actions the AI picks from while the program does the actual work.

why is my AI slower and more expensive since I turned connectors on?

Because a connector re-sends its whole list of actions on every single message, used or not. It is rent, not a per-use charge. One connector we used loaded about three hundred action descriptions on every message, so every thanks, one more thing re-sent all three hundred. The free fix is switching off the connectors you are not using in this session; nothing is lost.

is it safe to connect AI to my inventory system?

Start with looking only. A good connector keeps look and touch behind separate switches, so it can search and report long before it gets anything that creates, edits or deletes. When you do allow changes, keep the rule: it drafts, you approve. Ask for read-only logins and a log of every action it takes.

MCP in Plain English: Giving AI a Phone Line to Your Software

Questions this answersQuestions this answersfaq

what is MCP in plain English?what is MCP in plain English?what does an MCP server expose?: It stands for Model Context Protocol, and it is a standard plug. Without it an AI can only talk. With it, it can look things up in your software and, if you allow it, change them. The plug for one specific program is called a connector, and inside the connector is a list of actions the AI picks from while the program does the actual work.It stands for Model Context Protocol, and it is a standard plug. Without it an AI can only talk. With it, it can look things up in your software and, if you allow it, change them. The plug for one specific program is called a connector, and inside the connector is a list of actions the AI picks from while the program does the actual work.Model Context Protocol is a client-server protocol. A server exposes tools (callable, each with a name, description and input schema), resources (readable content addressed by URI) and prompts. The client lists them, passes the tool definitions into context, executes the structured call the model emits, and pastes the result back as text. Transports are stdio for a local child process and streamable HTTP for remote, with the credential held server-side.
why is my AI slower and more expensive since I turned connectors on?why is my AI slower and more expensive since I turned connectors on?what do tool definitions cost per turn?: Because a connector re-sends its whole list of actions on every single message, used or not. It is rent, not a per-use charge. One connector we used loaded about three hundred action descriptions on every message, so every thanks, one more thing re-sent all three hundred. The free fix is switching off the connectors you are not using in this session; nothing is lost.Because a connector re-sends its whole list of actions on every single message, used or not. It is rent, not a per-use charge. One connector we used loaded about three hundred action descriptions on every message, so every thanks, one more thing re-sent all three hundred. The free fix is switching off the connectors you are not using in this session; nothing is lost.Tool definitions are re-sent in full every turn, ahead of the message history. At roughly 150 tokens of schema each, forty tools is about 6K input tokens per turn before you type a word. Prefer servers exposing a curated core plus a search-and-load tool over ones that dump everything into tools/list, keep descriptions tight, and disable servers this task does not need.
is it safe to connect AI to my inventory system?is it safe to connect AI to my inventory system?what is the real risk of a connected tool surface?: Start with looking only. A good connector keeps look and touch behind separate switches, so it can search and report long before it gets anything that creates, edits or deletes. When you do allow changes, keep the rule: it drafts, you approve. Ask for read-only logins and a log of every action it takes.Start with looking only. A good connector keeps look and touch behind separate switches, so it can search and report long before it gets anything that creates, edits or deletes. When you do allow changes, keep the rule: it drafts, you approve. Ask for read-only logins and a log of every action it takes.Enable search, get and report tools first, with read-only credentials, per-tool allowlists, and audit logging of every call and its arguments; gate create and update behind confirmation. Know the boundary you are crossing: a tool result is untrusted text pasted straight into context, so a customer name or an order note can carry instructions the model cannot distinguish from yours. With read-only tools that is a nuisance; add a write tool or a URL fetcher to the same session and it becomes a path out.

Read the rest for freeSign in to read the restauth required past this point

A name and an email, once. The whole course is free, nothing is sold to you, and your place is kept so you can pick it up on your phone later.A name and an email, once. The course is free; the account exists so your progress and your level on the dial follow you between devices, and so you can be issued a certificate at the end.name + email. free. account carries completion state, data-level and certificate issuance.

We store your name, your email, which lessons you open and which you finish, and the reading level you use. Distru can see that. Nothing is charged and nothing is sold on.Stored: name, email, which lessons you open and complete, and your level on the dial. Distru staff can see it. No payment, no third-party advertising trackers.stored: name, email, per-lesson view + completion, data-level. visible to Distru. no payment, no ad-tech, no third-party pixels.