Most people use AI like a faster search engine: they ask it a question, get an answer, and move on. That's fine, but it leaves the biggest productivity gains on the table. The real leverage comes from using AI as a thinking partner — something that helps you triage an overflowing inbox, turn messy meeting transcripts into crisp action items, run a structured weekly review, draft the email you're dreading, and pressure-test a decision you keep avoiding.

The difference between a so-so AI session and a genuinely time-saving one is almost always the prompt. "Summarize this meeting" produces a vague paragraph. "Here is a transcript. Pull out every decision made, every action item with its owner and due date, and any open question that was not resolved. Format as three tables" produces something you can actually paste into your task manager. That gap — between vague instructions and structured ones — is what this guide aims to close.

Below are 20 prompt templates for the productivity tasks that eat the most time during a normal workweek: inbox triage, meeting notes, daily and weekly planning, email drafting, decision-making, research summarization, document review, and more. Each template is ready to copy, paste, and adapt. Use them in ChatGPT, Claude, Gemini, or any capable model — and keep your library versioned and reusable in Promptwright.

Why structured productivity prompts outperform generic ones

Before we get to the templates, it's worth understanding why structured prompts work so much better for productivity tasks. There are three reasons worth remembering.

First, constraints beat freedom. A prompt that says "make this shorter" gives the model a subjective target. A prompt that says "cut this to under 120 words, keep the three strongest arguments, drop all hedging language" gives the model a measurable one. Measurable constraints produce consistent, usable output.

Second, role and format shape output more than content. Telling a model "you are a meticulous chief of staff who values brevity" before asking it to summarize changes the tone, length, and emphasis of the summary far more than rewriting the question itself. Specifying format (bullet list, table, email draft) does the same.

Third, context is the multiplier. Productivity prompts live or die on the context you paste in: the actual transcript, the actual inbox list, the actual decision criteria. A great prompt with no context produces fluff. A decent prompt with rich context produces something genuinely useful. Always invest in pasting in the real material.

1. Inbox triage: turning an email dump into an action list

Most knowledge workers lose at least an hour a day to inbox anxiety — opening, half-reading, closing, reopening. The trick is to do a single bulk triage pass and then act. AI can do the triage for you if you give it the raw email list.

Prompt: Bulk inbox triage

You are a meticulous chief of staff. I will paste a list of email subjects and the first line or two of each email from my inbox, numbered. For each email, decide on a single suggested action and a priority (High / Medium / Low). Use these action categories:

- Reply (1-2 sentences) — needs a short reply, draft it
- Reply (longer) — needs a substantive reply, note the topic
- Read/review — informational, no reply needed
- Delegate — should be handled by someone else, note a suggested owner if obvious
- Archive — no action needed, can be filed
- Calendar — needs a meeting or task scheduled

Output a table with columns: #, Subject summary, Action, Priority, Suggested next step (one line). Sort by priority (High first). After the table, list the top 5 emails I should handle first.

Here are the emails:
[paste your numbered email list here]

This turns a wall of unread subject lines into a sequenced plan in under a minute. You can generate the input list by copy-pasting from your mail client or, if you use Gmail, exporting a quick list. Be sure to strip sensitive details if you're using a shared or non-logged-in AI tool.

2. Meeting transcripts into action items

The biggest waste in meetings isn't the meeting — it's the 20 minutes after, trying to reconstruct what was actually decided while it's still fresh. Hand the transcript to AI with the right structure and you get a usable summary in seconds.

Prompt: Decision-and-action summary

Act as a meticulous meeting note-taker. Below is a transcript of a meeting. Produce:

1. **One-paragraph summary** (max 4 sentences) of what the meeting was about and the headline outcome.
2. **Decisions made**: a bulleted list, each starting with "DECIDED:" followed by one sentence.
3. **Action items**: a table with columns Owner / Task / Due date / Notes. Pull owners only from what's explicitly said in the transcript; if an owner is unclear, write "Unassigned." If no due date was stated, write "TBD."
4. **Open questions**: anything that was raised but not resolved, as a bulleted list.
5. **Parking lot**: anything explicitly deferred to a future discussion.

Do not invent decisions or action items that aren't supported by the transcript. If unsure, leave that section empty.

Transcript:
[paste transcript]

This format is deliberately conservative — it tells the model not to invent. That's the key for meeting notes: you'd rather see an empty cell than a confidently fabricated action item.

3. Daily planning: a focused workday in 90 seconds

Open a model first thing in the morning, give it your calendar and tasks, and let it propose a realistic plan. The point isn't that the model knows better than you — it's that articulating your plan to a model forces clarity, and the model can sanity-check it against your real constraints.

Prompt: Daily plan with time blocking

You are my chief of staff. Today is [weekday, date]. My calendar:

[paste your calendar for the day, with times]

Tasks I'd like to make progress on (with rough size and importance):
1. [task] — [Small/Medium/Large] — [High/Med/Low importance]
2. ...

Constraints:
- I want a 30-min lunch break, away from screens.
- I work best in 60-90 min focus blocks and need 10-min breaks between.
- My most productive window is 9-11am.
- I should knock out my one Most Important Task (MIT) early if possible.

Propose a time-blocked schedule from [start] to [end] that:
- Puts the MIT in my most productive window.
- Groups related small tasks together (admin block).
- Leaves 15 min between focus blocks.
- Includes a 30-min lunch and two short breaks.
- Adds a 10-min end-of-day shutdown: a short list of what I accomplished and what carries to tomorrow.

Output as a simple time-blocked schedule in a table, then a 3-line note on any trade-offs you made.

Run this once and you'll do it every morning. The shutdown ritual at the end is what stops tasks from bleeding into your evening.

4. Weekly review: a structured reflection prompt

If you only adopt one prompt from this guide, make it this one. A 30-minute weekly review is the most reliably life-changing productivity habit most people never build — and AI makes it dramatically easier to actually sit down and do.

Prompt: End-of-week review

Act as a reflective chief of staff helping me run a weekly review. Below is a list of what I did this week (calendar entries, tasks completed, notes I took) and what I had planned. Help me think clearly.

Do the following, section by section:

1. **Headline wins** (3 bullets max): What were the most meaningful outcomes this week?
2. **What slipped and why** (bullets): Which of my planned items did not get done, and what got in the way?
3. **Energy patterns** (3-4 sentences): Based on when I did deep work vs admin, what does this say about my energy this week?
4. **One thing to stop**: One activity that was a clear waste of time or energy, with a 1-sentence suggestion to eliminate or delegate it.
5. **One thing to start**: One small habit or practice that would have made this week meaningfully better.
6. **Top 3 for next week**: My three Most Important Tasks for next week, with a one-line rationale each.

Be honest, not flattering. Where my data is ambiguous, ask me to fill in the gap rather than guess.

What I did this week:
[paste]

What I had planned:
[paste]

The honesty line matters. Productivity prompts tend to flatter — "great week!" — and that's worse than useless. Tell the model explicitly not to.

5. Drafting a difficult email

We all have emails we avoid because they're awkward, high-stakes, or emotionally charged. Sitting in front of a blank page makes them worse. Sitting in front of a decent first draft makes them survivable.

Prompt: Difficult email draft

Draft a professional email for me. Context:

- Recipient: [who, and our relationship]
- My goal for the email: [what outcome I want]
- The situation, briefly: [3-5 sentences of background, including any history]
- Tone I want: [firm / warm / apologetic / direct / diplomatic]
- Length: [short (~80 words) / medium (~150 words) / detailed]
- Constraints:
  - No corporate hedging language ("I hope this finds you well", "per my last email").
  - No filler ("just wanted to", "I wanted to reach out").
  - One clear ask, stated in the first two sentences.
  - If it could sound harsh, soften with one specific acknowledgment of the other side's perspective.

Give me the email body only, ready to paste.

The constraints are doing most of the work. The "no corporate hedging" and "one clear ask in the first two sentences" rules alone turn most AI drafts from simpering to usable.

6. Decision memo: pressure-testing a hard call

Big decisions — job offers, vendor choices, product direction — deserve more than a gut feeling. A structured decision memo forces you to articulate the question, constraints, options, and trade-offs. AI helps you draft it; then you make the call.

Prompt: Decision memo before committing

Act as a decision coach. Help me structure a decision memo before I commit to a path. Use this structure exactly:

**1. The decision** — one sentence, framed as a choice between named options.
**2. Why now** — what's forcing this decision by this date?
**3. Criteria** — what matters? List 4-7 weighted criteria (e.g., cost 25%, speed 25%, quality 30%, risk 20% — they should sum to 100%).
**4. Options** — 2-4 named options.
**5. Scorecard** — a table scoring each option against each criterion (1-5), with the weighted total.
**6. Risks per option** — top 2 risks for each option.
**7. Reversibility** — would this decision be easy, hard, or impossible to reverse if it went wrong?
**8. What would change my mind** — one piece of missing information that could flip my decision.

Here is my rough thinking on the decision:
[paste stream-of-consciousness notes, as messy as you like]

Help me fill in each section, especially criteria and the scorecard. Flag any of my reasoning that seems to skip a step or confuse options with criteria. Don't recommend a path — help me see what I'm actually choosing between.

The "don't recommend a path" instruction is important. You're using AI to see the decision clearly, not to outsource the judgment.

7. Long-form summarization with structure

The single most common productivity use of AI — "summarize this" — is also the one most often done badly. Generic summaries get vague. Structured prompts turn an article or report into something you actually use.

Prompt: 5-level structured summary

Below is an article/report. Produce a structured summary in five levels:

1. **TL;DR (1 sentence)** — the single most important takeaway.
2. **30-second version (50 words)** — the headline argument plus the key supporting point.
3. **3-minute version (< 200 words)** — the argument, the three strongest pieces of evidence, and the author's main recommendation, if any.
4. **Structure map** — one line per major section/claim, in order.
5. **Pull quotes** — the 3 sentences in the piece worth quoting verbatim, with brief context for each.

After the summary, list:
- **Terms or concepts a non-expert might not know**, each defined in 1 sentence.
- **Open questions the piece raises but doesn't answer**.
- **Disagreements or weak links** — any place where the argument feels overstated or unsupported.

Article:
[paste]

This is the difference between "AI summarized my report" and "I have a usable briefing on my report." The disagreement section is often where you learn the most — AI is a fast way to surface what's questionable in something you might otherwise have skimmed past.

8. Document contract / policy review

If you sign things (contracts, terms of service, vendor agreements, leases) without reading them closely, AI can do the close reading you don't have time for.

Prompt: Contract risk scan before signing

Act as a meticulous contract reviewer. Below is a contract / agreement. I am [my role, e.g., "an individual freelancer signing an SOW with a startup"]. Do NOT give legal advice, but do flag:

1. **Standard terms present** (1 line each): governing law, indemnity, IP assignment, payment terms, termination, confidentiality, liability cap.
2. **Unusual or missing terms**: anything that deviates from what I'd expect for this kind of agreement, or any of the above that's absent.
3. **Red flags**: clauses that could meaningfully work against me as [my role] — explain each in 2 sentences.
4. **Ambiguous language**: any sentence vague enough that interpretation is likely.
5. **Specific things I should confirm with a lawyer**, as a numbered list.

Contract:
[paste]

The model is not your lawyer, and this prompt is explicit about that. But it will consistently surface things you would have skipped — and give you a sharper list of questions to bring to an actual lawyer, which makes the paid hour more productive.

9. Project kickoff brief from notes

When you're starting something new, the first instinct is to start doing. The second instinct should be to brief it. AI can turn scratchy notes into a brief everyone can work from.

Prompt: Project kickoff brief

Turn my notes into a project kickoff brief. Use this structure:

- **Project name** (1 line)
- **One-sentence goal** (what success looks like, concretely)
- **Context** (2-4 sentences on why we're doing this now)
- **In scope / Out of scope** (bulleted lists)
- **Deliverables** (bulleted list, each with an owner placeholder)
- **Timeline** (milestones, best guess dates)
- **Open questions to resolve kickoff** (bulleted list)
- **Risks and dependencies** (bulleted list)
- **Definition of done** (2-3 sentences)

Notes:
[paste your messy brainstorm]

Match the tone of a senior PM writing a brief for the team to self-organize from. Don't invent dates or owners — use placeholders like [TBD] where unknown.

Paste the output into your team's docs tool and you have a real brief in 5 minutes that would have taken an hour to write from scratch.

10. Quick research on a topic I need to be conversant in

We all have moments where we need to sound informed about something fast — a new regulation, a tool, a market category, a competitor. The trick is to ask AI to give you the shape of the topic, not a comprehensive summary.

Prompt: Crash course on a topic

Give me a crash course on [TOPIC] that I can read in under 5 minutes and walk away conversant. Use this structure:

1. **What it is, in plain language** (2 sentences).
2. **Why it matters right now** (2 sentences).
3. **The 5 most important subtopics or terms** — each as a one-line definition.
4. **The 2-3 main camps / viewpoints on it**, if any — what they argue, in one sentence each.
5. **A common misconception** about it (one sentence, with a one-line correction).
6. **3 good questions to ask someone knowledgeable about it** — questions that show you've thought about it.
7. **3 reputable sources for going deeper** — name them and the type of source (book, report, podcast, etc.).

Be specific and concrete. No filler.

This works because the questions section gives you a way to engage — most people who "know" a topic just know how to ask questions about it. Good questions buy you credibility while you keep learning.

11. Status update generator

Status updates are pure process — they're important, they're repetitive, and everyone hates writing them. Perfect AI candidate.

Prompt: Weekly status update from my tasks

Below is my task list and notes from this week. Write a status update for [audience: team / manager / client] covering this week and previewing next week. Use this format:

**This week:**
- 3-5 bullets on completed work, ordered by importance to the audience.
- 1-2 bullets on in-progress work, with one line on status (on track / blocked / risk).
- 1 line on blockers, if any — and a concrete ask where applicable.

**Next week:**
- 3-5 bullets on planned work, ordered by importance.

**Decisions / asks:**
- 1-3 bullets on decisions I need from the audience or topics I want their input on.

Tone: direct, no corporate filler, no hedging. Sentences are short. Verbs are specific (e.g., "shipped", "revised", "debugged", not "worked on").

Tasks / notes:
[paste]

If you do this every Friday, status updates go from a 25-minute chore to a 3-minute one. Over a year, that's a full workday back.

12. Learning from a mistake: structured post-mortem

Personal post-mortems, done well, are one of the highest-leverage things you can do — they extract lessons from experiences you'd otherwise just have to repeat. But they require honesty, which is uncomfortable. AI makes the structure easy and the honesty less painful.

Prompt: Personal post-mortem

Help me run a blameless post-mortem on a recent outcome I want to learn from. Structure:

- **What happened** (2-3 sentences, factual, no judgment yet)
- **What I expected** (1-2 sentences)
- **The gap** (1-2 sentences)
- **What I knew then vs what I know now** — the key new information that wasn't available at the time
- **Decisions I made that, with the information I had THEN, were / were not reasonable** (separate lists)
- **System-level lessons** — what process, structure, or default I could change so this is less likely to recur
- **One concrete next action** to lock in the lesson

Be concise. Don't moralize. If I start slipping into blame (self or others), flag it and restate the point as a system question.

Situation:
[paste rough notes]

The "blameless" frame comes from engineering post-mortems, and it works on personal mistakes too. The whole point is to extract the lesson and the change, not to relitigate the decision.

13. Negotiation prep

Before any important negotiation — a salary conversation, a vendor contract, a scope-of-work discussion — prep matters more than tactics in the moment. AI can run the prep with you.

Prompt: Negotiation prep brief

Help me prep for a negotiation. The counterpart is [who], the subject is [what], and the stakes for me are: [brief].

Produce a brief covering:

1. **My interests** (not positions) — list 3-5 underlying needs I have (e.g., "more flexibility", "higher base", "less weekend work").
2. **Their likely interests** — 3-5 needs they probably have, based on incentives and role.
3. **ZOPA guess** — what's the plausible zone of agreement here, in plain terms?
4. **BATNA** — my best alternative if we don't reach an agreement. Strengthen it now if I can.
5. **Three opening framings** — three ways to open the conversation that emphasize shared interest.
6. **Three possible concessions / trades** I could offer that cost me little but might mean a lot to them.
7. **Two questions to ask the counterpart** that would reveal their real constraints.
8. **Walkaway point** — at what point do I leave the table, and what do I do next?

Be honest about walkaway options even if uncomfortable.

Coming in with a BATNA and ZOPA articulated — even roughly — changes how you sound in the room.

14. Reading list prioritization

We all have a pile of "want to read" that grows faster than we can read it. The honest answer is we'll never read most of it. AI can help you triage.

Prompt: Reading list triage

Below is my reading list. For each item, decide:

- **Why I probably added it** (1 line, best guess from title/author)
- **Likely value to me** (High/Medium/Low) given that my current focus is [your focus, e.g., "scaling a solo consulting practice"]
- **Suggested action**: Read now / Skim only / Read someday / Drop — for each, a one-line rationale.
- For the top 3 to "Read now", suggest whether to read the whole thing or just a specific chapter.

Then give me a single short-listed priority order (top 5) for what to actually start next.

Reading list:
[paste]

Most reading lists shrink to a manageable three or four after this. The value isn't just saving time — it's making reading feel like a choice again, not a backlog.

15. Brainstorming partner for anything

Sometimes you just need to think out loud with someone who won't get tired. AI is shockingly good at being a useful brainstorming partner if you give it structure.

Prompt: Structured brainstorm

I'm brainstorming [TOPIC]. My rough starting ideas (if any): [paste].

Run a structured brainstorm:

1. **Restate the problem in one sentence**, sharper than I did.
2. **5 direct variations** on my existing ideas — small tweaks that might matter.
3. **5 lateral ideas** that draw from unrelated fields or angles I might not have considered.
4. **3 contrarian ideas** — ideas that might seem wrong but have a kernel of insight.
5. **2 "obvious but I haven't said it" ideas** that I might be avoiding for some reason.
6. **One organizing insight**: what do my good ideas have in common?

Number everything. One line each — short enough for me to scan and react to, not paragraphs.

The "obvious but I haven't said it" section regularly surfaces the thing I actually needed to consider.

16. Turning a wall of notes into an outline

For bigger writing projects, the hardest jump is from a pile of notes to an outline. Notes are stream-of-consciousness; outlines are sequenced argument. AI does that translation well.

Prompt: Notes-to-outline

Below is a mess of notes I've collected on [TOPIC] — quotes, half-formed ideas, examples, reactions, links. I want to turn them into an outline for a [blog post / essay / report / chapter] of roughly [target length].

Produce:

1. **Possible thesis statements** — 3 options, one sentence each, with different angles.
2. **A working outline** supporting the strongest thesis:
   - Major sections (H2-level), each with:
     - A one-line description of what this section argues
     - 2-4 sub-points (H3-level)
     - Which of my notes/quotes/examples belong here, by short snippet
3. **Notes that don't fit** — flag anything in my notes that didn't find a home, with a one-line suggestion to either cut it or create a section for it.
4. **One suggestion** for what would make the piece stronger that I'm missing.

Notes:
[paste]

Starting a writing project from an outline (not blank page, not a pile of notes) is the difference between writing and staring at a screen.

17. Quick critique of a draft

We've all sent the email / posted the thing / hit publish without enough review time. AI does fast critique well — if you ask for the right kind.

Prompt: Pre-send critique

Critique the following draft. Be specific and unsentimental. Cover in order:

1. **Opening**: Does the first sentence do real work? If not, suggest a sharper one.
2. **Structure**: Is there a clear throughline? If you had to write the throughline in one sentence, what is it? If you can't, say so.
3. **Clarity**: 2-3 sentences that are unclear or could mean two things. Quote each and restate what you think the author meant.
4. **Length**: 1-2 places that could be cut by 30%+ without losing meaning. Quote them.
5. **Tone**: 1-2 places where the tone wobbles (gets too casual or too stiff). Quote them.
6. **Persuasiveness**: Where the argument is weakest. One line each.
7. **One thing to add** that would meaningfully strengthen it.

Do not change my draft or rewrite it. Quote and critique only. After the critique, give me one overall verdict sentence.

Draft:
[paste]

Most AI critique prompts ask the model to "rewrite this for clarity," which strips voice and produces forgettable prose. Critique-only preserves your voice and improves your drafts.

18. Turning decisions into written rationale

A surprising amount of work friction comes from undocumented decisions. Six months later, no one remembers why something was decided. AI turns your "we decided to do X because Y" notes into a memo in 30 seconds.

Prompt: Decision rationale memo

Draft a decision rationale memo (Architecture Decision Record style) for the following decision. Use this structure:

- **Context** (1 paragraph: what situation led to this decision)
- **Decision** (1 paragraph: what we decided)
- **Status**: Proposed / Accepted / Superseded
- **Consequences** — positive, negative, neutral — of this decision
- **Alternatives considered** — 1-2 sentences each, with why not
- **Open questions** this decision raises

Write in factual, low-drama style, like a competent staff engineer documenting a decision for a future reader who didn't attend the meeting.

My rough notes on the decision:
[paste]

These memos compound. A team that does them has a record; a solo operator who does them stops second-guessing their own past decisions.

19. Weekly energy audit

Productivity isn't just about what you do — it's about whether you have the energy to do it. Most productivity advice skips energy entirely. AI can do a light energy audit if you track even rough data.

Prompt: Energy audit from a week of notes

Below is a rough log of my past week — what I did, when, and a one-line note on how I felt (energy level / mood / focus) where I have it. Help me see patterns.

Output:

1. **My peak energy windows** — the times of day where my best focus shows up most consistently.
2. **My trough windows** — where my effort consistently drops.
3. **Activity-energy fit** — mismatches I made this week between activity type and energy level (e.g., deep work during a trough, easy admin during a peak).
4. **Drains** — activities that consistently correlate with energy drops (even if I don't enjoy admitting it).
5. **Replenishers** — activities that consistently correlated with energy recovery.
6. **Two concrete shifts** for next week based on these patterns.

Be skeptical of my self-reporting where it's inconsistent with my actions.

Log:
[paste]

This is the one most people skip, and the one that produces the most durable insights over time.

20. End-of-day shutdown ritual

Closely related to the morning plan — a 5-minute shutdown at the end of the day stops work from leaking into your evening. AI can prompt and capture it.

Prompt: End-of-day shutdown

Help me run a 5-minute end-of-day shutdown. Below is what I did today, including things I started but didn't finish (and the state they're in). Ask me, one section at a time, and let me respond before going on:

1. What did I actually accomplish today? (Capture as a short bulleted list — done is done.)
2. What's still open and unfinished? (For each: do I want to hand it off tomorrow morning, defer it, or drop it?)
3. What's the single most important thing I want to start with tomorrow? (One MIT.)
4. Is there anything still on my mind I haven't captured? (Brain dump — I'll list it.)
5. Based on the above, draft tomorrow's morning first-three list and a one-line note about how today's work connects to next week's MIT.

At the end, write a 2-sentence "shutdown statement" I can read aloud: an acknowledgement of the day's work and a clear separation between work-time and evening-time.

Today's notes:
[paste]

The shutdown statement sounds silly to read aloud, and is surprisingly effective at actually stopping your brain from running laps over unfinished threads.

How to make these prompts stick

Templates only help if you actually use them. The friction is not writing prompts — it's hunting for the right prompt when you need it.\n\nThree habits to install:

Start every day with two or three of them. The morning plan and end-of-day shutdown should become reflexive. Don't rely on willpower; create a shortcut or a one-line index card with the prompt names and stick it on your monitor for the first month.
Keep a small library, not a big one. Twelve prompts you actually use beat a hundred you never open. Curate. Cut the ones you've stopped running.
Iterate when a prompt doesn't work. When AI returns something off, the instinct is to blame the model. Usually the prompt is missing a constraint. Add the missing constraint, save the updated version, and re-run. Over a month, your prompts get better even if you don't get smarter.

A tool like Promptwright is built for exactly this: keep your prompt library versioned, organized, shareable across a team, and run them against your preferred models without re-typing.

Common mistakes to avoid

A few patterns that derail even structured prompts:

Pasting the prompt without pasting the data. Most productivity prompts need real input — a transcript, a task list, a draft. Without it, the model has nothing to act on.
Asking for "a summary" with no shape. Always specify format (table, bullets, paragraph), length, and what to emphasize.
Treating AI output as truth. For meeting summaries and decision memos, treat AI drafts as a first pass you verify. The model will sometimes invent an action item — the prompt above explicitly tells it not to, but you still verify the important ones.
Using AI for energy management. The prompts here can help you see your energy patterns, but they can't fix a 4-hour-sleep week. Respect the body.
Letting prompts pile up unused. A graveyard of abandoned prompts is worse than no prompts. Curate.

Ready to use these prompts?

The 20 templates above are designed to be copy-pastable on day one. The bigger win comes when you stop copy-pasting them ad hoc and start keeping them somewhere structured — where you can revise them, share them with teammates, and run them in one place.

If you'd like a home for your productivity prompts, Promptwright gives you a versioned, team-shareable prompt library and lets you run your prompts against leading models without the back-and-forth. It's a small change that compounds — every prompt you refine once and reuse saves you the rewriting cost every time after.

Start with the morning plan and the weekly review. Those two, run consistently, will give you back the hours the rest of these prompts help you use better.

AI Prompts for Productivity: 20 Templates to Reclaim Your Workday