Capture recurring AI tasks as reusable templates — with variables, steps, and a fixed output format.
A workflow is a saved prompt template with variables, optionally several steps, and a fixed output format. Instead of retyping the same prompt every Monday, you fill in the input fields and hit Run.
20 minutes of copy-paste prompting turn into 30 seconds of filling out a form — if the template is built cleanly. This page shows what cleanly built means.
Variable types: short text, text block, dropdown, on/off
Per-step model choice
Per-step web search toggle (useful when only research steps need live information)
Pinned reference documents (always attached) plus optional per-run file upload
Output formats: Markdown in chat, Word/PDF/text, Excel/CSV, PowerPoint, structured JSON (for downstream processing in Make/n8n or your own APIs)
Share with a project → colleagues run the workflow without editing it
Note
PowerPoint and Excel as a fixed file output (e.g. for a recurring, consistently formatted report) work via workflows with the document output toggle enabled. In a normal chat, an assistant can also generate or edit .pptx or .xlsx files - using the Code Sandbox or a Skill instead of a fixed output format. For a recurring process with a guaranteed consistent format, build a workflow; for a one-off or variable request, a normal chat is enough.
Note
Skills and the Code Sandbox only run in an assistant, not in a workflow. If your recurring process needs custom code — e.g. a script that fills a Word template or calls an external API with no native connector — build it as a Skill in an assistant. A workflow step cannot mount a skill.
How to build a workflow step by step, run it, and organize it — plus the mistakes to avoid along the way — lives on its own page: Create and Run Workflows.