A session is one continuous working conversation with an AI tool: the active room where work happens. In Claude Code or Codex it includes your instructions, the files the agent has read, the tool results, the mistakes it corrected, and your decisions. Every session has a context window, so its working memory is finite.
A working meeting captures it. You sit down, put the relevant papers on the table, talk through the job, and produce work. If it runs six hours, covers seven unrelated topics, and nobody writes down what changed, the meeting gets messy.
Why you care
Use sessions to hold phases of work: one for research, another for drafting. Ask one session to write copy, debug code, review a contract, and plan a training, and you’re asking one meeting to hold five. When a phase is done, close it cleanly and hand the next session the useful output rather than the whole conversation; when a session fills up, compaction summarizes it and details can drop. At the end, ship the output or turn the useful context into something durable: a note, a file, a skill. The quality of the meeting shapes the quality of the work.