Glossary
Agentic workflow
An agentic workflow is a process where an AI agent works toward a goal in several steps: it plans what to do, uses tools such as search, documents or other software, checks what it got and decides the next step, instead of answering a single prompt once.
Examples of agentic workflows
An agent researches a list of companies, opens their sites, fills a table and flags the ones that fit. Another reads incoming invoices, matches them to orders and prepares the unclear ones for a person. Each runs several steps with tools, where a chat would give one answer.
Agentic workflow vs multi-agent system
An agentic workflow can use one agent or several. A multi-agent system splits the work between agents with different roles, for example one that researches and one that checks. More agents add coordination, not quality by themselves; what matters is the context they share and how results are checked.
When a business is ready for agents
Agents work best when they plug into a system that already works: shared knowledge, tasks with clear criteria and checked results. An agent at the first stage is a new employee nobody briefed. That is why agents come after the four stages of AI maturity, not instead of them.