Glossary

AI Agent

Agent

An AI agent is software that makes a plan for a job and then uses tools to carry it out, instead of only answering a question in text. On IBM i that is the difference between a chatbot that explains RPG and IBM Bob, which IBM says you connect directly to IBM i to read and write source, compile programs and run SQL.

What it isSoftware that plans a job and then calls tools to do it, instead of only answering in text
IBM's definitionA system that autonomously performs tasks by designing workflows with available tools
Runs on IBM i?IBM's AI on IBM i guide marks Runs on IBM i with a no for watsonx Orchestrate; agents reach the platform through tools
Who sets the goalA person. IBM says agents require goals and predefined rules defined by humans

What is an AI agent?

Think about two helpers at a travel desk. Ask the first one which flight gets you to Chicago on Tuesday and it reads you the answer. Ask the second one to get you to Chicago on Tuesday and it checks your calendar, picks the flight, books it and puts the booking back on your calendar. The first one talks. The second one acts. An AI agent is the second one.

IBM's own wording is that an AI agent is a system that autonomously performs tasks by designing workflows with available tools. The word tools is the important one. A tool is anything the agent is allowed to call: a database query, a REST call, a command. Take the tools away and the agent is back to talking.

  • Tool calling. IBM says agents use tool calling on the backend to obtain up-to-date information, optimize workflows and create subtasks autonomously.
  • Memory. IBM describes this as the agent's ability to store past interactions in memory.
  • Planning. IBM says the agent creates a plan of specific tasks and subtasks to accomplish the complex goal.
  • Reasoning. IBM describes agentic reasoning as continuously reassessing its plan of action and making self-corrections.

How is an agent different from a chatbot?

A chatbot answers, an agent acts

QuestionNonagentic chatbotAI agent
What it does with your requestReturns an answer in textMakes a plan, then runs the steps
Tools it can callNone, in IBM's descriptionWhatever tools it has been given
Memory of past workNone, in IBM's descriptionStores past interactions
How far ahead it can workShort-term goals only, and it cannot plan aheadBreaks a goal into tasks and subtasks
What it can touch on IBM iNothing. It only talks about your codeOnly what its tools expose, such as SQL through the IBM i MCP Server

The chatbot column uses IBM's wording for nonagentic AI chatbots: ones without available tools, memory or reasoning, which can reach only short-term goals and cannot plan ahead. Read from IBM's AI agents topic page on November 19, 2026. MCP stands for Model Context Protocol, an open standard for connecting AI tools to outside data and actions. Table rows are not linked because this site's content blocks render as plain text.

How does an agent reach IBM i?

The three ways IBM's AI on IBM i guide lists

  1. Wrap your own business functions as REST toolsIBM's first option is to expose IBM i business functions as OpenAPI-defined tools. OpenAPI is a standard way to describe a web service so other software knows how to call it. The agent then calls your program the same way any other app would.
  2. Point the agent at the IBM i MCP ServerIBM's guide says the IBM i MCP Server gives native access to SQL, IFS, CL commands and system objects. IFS is the Integrated File System and CL is Control Language, IBM i's command language. The project's own page says it enables AI agents to interact with IBM i systems through the Model Context Protocol and gives secure, SQL-based access to Db2 for i, using Mapepire for the database connection.
  3. Register an agent you built yourselfIBM's third option is Agent Connect, which it describes as a way to register custom IBM i automation agents built on any framework. That is the door for a shop that already wrote its own automation and wants it managed with the rest.

Read from the watsonx Orchestrate page of IBM's AI on IBM i guide on November 19, 2026. The same page marks Runs on IBM i with a no, marks on premises and IBM Cloud with a yes, and says all options support hybrid scenarios where IBM i workloads stay on premises while orchestration is managed centrally.

What does an agent do on IBM i today?

The clearest working example is coding. IBM's page for IBM Bob on IBM i says it is built for RPG, CL, DDS, SQL, COBOL and Db2 for i environments, and that you connect Bob directly to IBM i to read and write source, compile programs, run SQL and support testing from within the development workflow. Reading source is chatbot work. Writing source and compiling a program is agent work.

IBM adds on the same page that Bob helps teams apply repeatable modernization patterns while keeping developers in control, and IBM's agents topic page says agents require goals and predefined rules defined by humans. Both are vendor statements about how the product is meant to be used, not a technical limit. The practical control in your shop is the tool list and the user profile behind it: an agent can only do what its tools let it do, and it does that with the authority it signs in with. The IBM i MCP Server project defines its SQL tools in YAML files, so the list of tools is something you decide and can read. YAML is a plain-text file format. The project's page showed a Status: Stable badge and named MCP spec version 2026-07-28 when it was read on November 19, 2026.

Related Terms

IBM Bob (Project Bob)

IBM's generally available AI coding agent for IBM i, built as a fork of Visual Studio Code and tuned for RPG, CL, DDS, and SQL. Reached GA on March 24, 2026; Bob 2.0 and the IBM i-specific Bob Premium Package for i shipped June 24, 2026.

Model Context Protocol

An open standard, released by Anthropic on November 25, 2024, that defines a client-server architecture for connecting AI applications to external data sources, tools, and prompts over JSON-RPC 2.0. IBM's IBM i MCP Server is an implementation of it for Db2 for i and other IBM i resources.

IBM i MCP Server

An open-source, IBM-published server implementing the Model Context Protocol (MCP), letting AI agents and LLMs query Db2 for i data and take defined actions on IBM i using natural language. In beta since October 2025.

watsonx Orchestrate

IBM's AI agent and automation platform for orchestrating multi-step business workflows across enterprise applications. Since August 2026 it can launch IBM Bob from its Create an agent dialog through a Build with Bob preview, but it still ships no IBM i or Db2 for i connector.

watsonx Orchestrate Agent Builder

watsonx Orchestrate Agent Builder is where an AI agent is actually defined: its name, its written instructions, the large language model that reasons for it, and the list of tools it is allowed to call. For an IBM i shop the tool list is the whole story, because it accepts a Model Context Protocol (MCP) toolkit, and that is the practical way to point an Orchestrate agent at Db2 for i data when IBM ships no IBM i connector.

Agent2Agent Protocol

An open protocol that lets one AI agent hand work to another agent it did not build, so an IBM i shop can keep its Db2 for i tools behind its own agent and still let an outside agent ask that agent for an answer. Google published it in April 2025 and gave it to the Linux Foundation on June 23, 2025; IBM's watsonx Orchestrate connects external A2A agents, and IBM's ContextForge MCP Gateway federates A2A servers beside Model Context Protocol servers.

IBM Agent Identity

Gives an AI agent its own registered identity, a named owner and short-lived access tokens, so a shop can see which agent asked for data instead of finding a borrowed human login in the audit trail. IBM made it generally available on October 26, 2026 at IBM TechXchange 2026, and it runs with IBM watsonx Orchestrate, IBM Verify or Microsoft Entra, and HashiCorp Vault.

Large Language Model (LLM)

A machine learning model trained on large volumes of text, used as the underlying technology behind generative AI tools including IBM Bob and general-purpose assistants like ChatGPT and Claude.

Generative AI

AI systems that generate new content, such as code, text, or explanations, rather than only classifying or scoring existing data.