Prompt templates for everyday work tasks

What is a prompt template?

A prompt template is a prompt written once with placeholders, such as [reader] or [paste notes here], so the same structure can be reused for many similar tasks. The fixed parts carry the instructions that rarely change; the blanks carry the facts that always do.

The term also has a technical meaning. In developer frameworks, a prompt template is a string in code with variables that a program fills automatically. This guide is about the everyday kind: text that a person copies into ChatGPT, Claude, Gemini or Copilot and completes by hand. The principles overlap, but you do not need any code to use these.

What is the difference between a prompt template and a prompt framework?

A framework is a checklist of what every prompt should contain; a template is a specific prompt built from that checklist for one kind of task. Frameworks help you write a new prompt; templates help you repeat a good one.

You do not need a branded acronym. Microsoft's Copilot guidance uses goal, context, source and expectations. OpenAI Academy uses task, background and desired output. The framework below merges those into one line per element, and every template on this page follows it.

What does a good general-purpose template look like?

A general-purpose template states the five elements in order and ends with an instruction to list gaps instead of filling them. Copy this and replace the bracketed parts:

Task: [Draft / summarise / compare / rewrite] [the deliverable, with length]. Reader: [who], who needs to [decision or action]. Source: use only the material below. If something important is not in it, do not guess; list it under Questions at the end. Limits: [tone]; keep these facts unchanged: [facts]; do not add [results, dates, names, prices or other claims]. Format: [structure]. Source material: [paste here].

The final instruction does more work than it looks. Without it, most assistants fill gaps with plausible material. With it, the gaps arrive as a short list you can answer or pass to someone who knows.

Which prompt templates are most useful at work?

Six templates cover a large share of routine office requests: summarising, drafting a message, comparing options, extracting data, rewriting for a new reader and generating ideas within limits. Each is shown with its blanks and the check to run afterwards.

How do you fill in a template? A worked example

Fill the blanks with specific facts, not with more instructions. Here is the summary template used by a fictional 25-person design agency, Copperleaf Studio, whose operations lead must decide whether to renew a freelance contract.

Filled prompt: Summarise the material below for the operations lead, who needs to decide whether to renew a freelance illustrator's contract for next quarter. Give the decision-relevant points first, then open questions, in under 200 words. Quote figures exactly as they appear and do not infer causes the material does not state. Source material: [three project notes, the current contract terms and two emails from account managers].

A typical weak result, before the limits were added, said the illustrator had been consistently late, which made the decision look obvious. The notes mentioned one late delivery, and one account manager's email said that delay followed a late brief from the client. The reviewed summary reads: one delivery was late; one account manager attributes the delay to a late client brief; the other projects were delivered on the agreed dates. Open question: is the day rate in the contract still current? That version gives the operations lead the facts and leaves the judgement with them.

How should a team keep a shared prompt library?

Keep a small library with a named owner, a date on each template and a note on which approved tools it is meant for. A large, unowned collection of copied prompts quickly fills with stale facts and duplicated versions.

Some tools have built-in saving. Microsoft 365 Copilot lets users save a prompt that has run successfully and view it later; Microsoft notes that saved prompts are visible only to the person who saved them, while Microsoft-curated prompts are visible across the organisation. Check what your own tools offer before building a separate system. A shared document with a short table of contents is enough for most small teams.

When do prompt templates go wrong?

Templates go wrong when they replace thinking about the specific task. A template filled with vague answers to its blanks produces a vague result with a confident structure, which can be worse than an obviously rough draft.

They also fail when the task is unusual. A complaint from a long-standing client, a sensitive staff message or anything with legal or financial consequences deserves a prompt written for that situation and a qualified reviewer, not a reused template. And no template makes an AI tool suitable for data your company has not approved it to handle.

Finally, be wary of large downloadable packs of hundreds of prompts. They can be a useful source of ideas, but most entries assume a generic business and omit the source and limits that make a prompt reviewable. Five templates your team has tested on its own work are worth more than a long list nobody has checked.

How can a team practise using templates well?

Practise by giving two people the same template and the same fictional source, then comparing what each filled in and what each output got wrong. The differences show which blanks need clearer guidance.

Tosheo.ai is planned to use this kind of exercise inside role-specific lessons: a short explanation, a worked example from the learner's role, then a hands-on task with a real AI tool, scored against a visible rubric. It is pre-build, so treat this as the intended design. The prompt engineering guide covers the underlying principles, and the ChatGPT and Copilot prompt guides apply these templates to specific tools.

Sources and scope

Microsoft Support, Get started writing prompts in Microsoft Copilot, accessed 16 October 2026 ↗

OpenAI Academy, Prompting resource, accessed 16 October 2026 ↗

Microsoft Support, Saving prompts to use again in Microsoft 365 Copilot, accessed 16 October 2026 ↗

Anthropic, Claude docs: Prompting best practices, accessed 16 October 2026 ↗