What is a prompt template (and prompt patterns)?

A prompt template is a reusable prompt with blanks you fill in — the fixed wording stays the same and you swap in the variable parts each time. "Summarize the following {document} in {number} bullet points" is a template. Prompt patterns are named, repeatable structures — like giving the model a persona or pinning down the output format — that reliably steer a response a certain way.

Why do teams use templates?

Templates turn a prompt that worked once into something anyone can reuse without guesswork. Instead of every person rewriting the same request slightly differently, you standardize on one wording that has been tested, and you parameterize the parts that change. That means consistent output, easier debugging when something breaks, and a prompt you can improve in one place for everybody.

It is the difference between a one-off message and a small piece of reusable software.

What are some common prompt patterns?

A few patterns show up again and again:

• Persona — tell the model who to be: "Act as a patient math tutor." Sets tone, depth, and vocabulary in one line.
• Output-format — specify the exact shape of the answer: a table, JSON, three ranked options, a single sentence.
• Flipped-interaction — have the model interview you: "Ask me questions one at a time until you have enough to write the brief." Useful when you don't know everything up front.
• Template — hand the model a fill-in-the-blank structure and tell it to keep the placeholders exactly where they are.
• Recipe — you give the goal and known steps, and the model fills in the missing steps in order.

How do you start using them?

Take a prompt you already trust, mark the parts that change with clear placeholders, and save it where your team can find it. Layer patterns as needed — a persona plus a strict output-format covers a surprising number of jobs. Keep a shared library and note which template is for what.

Over time you stop writing prompts from scratch and start assembling them from parts you have already proven work.