OpenAI’s current Dots overview makes a distinction that matters when turning a helpful conversation into recurring work. It says conversations with a dot do not count toward ChatGPT usage limits, while tasks the dot starts or manages in Work or Codex use those products’ allowances. The plan also includes an allowance for deeper work. A long chat and a delegated build therefore should not be treated as interchangeable measures of usage.
The guide describes extended limits for the first month after launch, but its usage paragraph does not give a numeric deeper-work quota or an exact transition timestamp. This is an analysis of current documentation, not an announcement of new pricing or a prediction of a particular account’s remaining capacity. The page has no displayed absolute publication or revision date.
Editorial take: Judge an agent workflow by completed, useful outcomes and the allowance it consumes. Before adding another daily routine, identify whether it mainly chats, researches, or delegates substantial work. Track which jobs finish and which hit a limit; counting messages alone can hide the work that determines capacity.