📈
AI for Marketing
Marketers ship more, faster — but generic AI output is everywhere. The win is making your AI sound like your brand, not the model. These prompts teach the specificity and constraint patterns that produce on-brand work.
Best tools
ChatGPT
Claude
Perplexity
Punchy ad headline (with constraints)
Write 10 ad headlines for a B2B SaaS that helps marketing teams cut campaign reporting from 6 hours to 20 minutes.
Constraints: under 9 words, no emojis, no questions, no "transform" / "unlock" / "revolutionize". Each headline should name a specific pain or benefit.
Audience: VP of Marketing at a 50-200 person B2B company. Tired of vendor jargon.
Try in playground →
SEO brief from a SERP
I'm writing a blog post targeting the keyword "best AI for sales emails". Here are the titles of the top 10 results: [paste].
Tell me:
1. What angle do most articles miss?
2. What's the strongest unique angle for our post?
3. Which 3 questions would the reader want answered that the current top results don't address well?
4. Suggest an outline (H2s only).
Try in playground →
Campaign retro analyst
Here's our campaign data: [paste CSV: channel, spend, impressions, clicks, conversions, revenue].
Act as a marketing analyst. Tell me:
- Which channel had the best ROAS, and by how much
- Which channel underperformed and the most likely reason
- One specific test I should run next month to validate your hypothesis
- The number I should track to know if it worked
Be direct. Skip the executive summary.
Try in playground →
LinkedIn post in your voice
Here are 5 LinkedIn posts I've written that performed well: [paste].
Write a new LinkedIn post about [topic] that sounds like the same person wrote it. Match my sentence rhythm, the way I open posts, and the way I use line breaks. Don't add emojis I wouldn't use. Don't end with a question unless I usually do.
Try in playground →
Full track coming Q3 2026: Brand-voice training across all your AIs, the campaign retro playbook, SEO content systems that don't get penalized, and a full LinkedIn writing curriculum.
💼
AI for Sales
Sales is where bad AI shows up loudest — generic outreach, "circling back" emails, soulless follow-ups. The wins here come from specificity: referencing actual demo content, mirroring the prospect's language, and pruning every word that doesn't move the deal.
Best tools
Claude
ChatGPT
Perplexity
Follow-up after a demo (specific friction)
Write a follow-up email to [name] at [company]. We demo'd 9 days ago — went well, no objections, then total silence.
At the end of the demo they said: "[paste exact quote about a pain point]"
The email should reference that exact pain point, propose one specific 15-minute next step, and give them an easy out. Under 80 words. No "circling back". No "just checking in".
Try in playground →
Pre-call account research
Research [company name]. I'm meeting with their [title] tomorrow. Give me:
1. What does this company do, in one sentence?
2. Recent news (last 6 months) — funding, leadership changes, product launches, layoffs
3. Their stated strategy from their most recent earnings call or investor letter (if public) or About page (if not)
4. Three smart questions I could ask that would show I did my homework
5. Two specific topics to avoid based on what I found
Cite sources.
Try in playground →
Cold email — one specific frame
I'm selling [product] to [persona] at [company size]. Write a cold email built around ONE specific frame: I noticed that [observable thing about their company / industry / role] — here's a way [product] addresses it.
Rules: 60 words max. No "Hope you're well". No "I'd love to". One question at the end. Subject line under 6 words.
Try in playground →
Call notes → CRM update
Here are my raw notes from a sales call: [paste].
Turn this into:
1. A 4-sentence summary for my CRM
2. Next steps (with owner and date)
3. Three quotes from the prospect I should remember verbatim
4. One risk to the deal I should flag to my manager
Try in playground →
Full track coming Q3 2026: The complete sales follow-up curriculum, prospect-research playbooks, objection-handling scripts, and call-notes-to-CRM automation patterns.
⚙️
AI for Engineers
Engineers don't need AI to write boilerplate — they need AI to be a sparring partner. Architecture decisions, code review, debugging dead ends, and explaining unfamiliar codebases. These prompts treat the AI like a senior staff engineer, not an intern.
Best tools
Claude
ChatGPT
Architecture sparring partner
I'm designing [system]. My current plan is [paste design or diagram description].
Act as a senior staff engineer doing a design review. Don't agree with me. Push on:
- What breaks at 10x scale?
- What's the simplest version that ships next month?
- What load-bearing assumption am I making that might be wrong?
- What's the one thing a future engineer will hate about this?
Be specific. No platitudes.
Try in playground →
Debug a perplexing bug
I have a bug. Here's what I'm seeing: [paste behavior].
Here's the relevant code: [paste].
Here are the things I've already ruled out: [list].
Don't just suggest random fixes. Walk through the 3 most likely root causes in order of probability, what would prove or disprove each, and which one I should investigate first based on what's fastest to verify.
Try in playground →
Explain unfamiliar code
Here's a function I need to understand: [paste].
Explain it in three passes:
1. What does it do, in one sentence, like I'm a product manager.
2. How it works, walking through the control flow line-by-line.
3. What's subtle or non-obvious about it — bugs that could happen, edge cases, performance characteristics.
Try in playground →
Code review on yourself
Here's a PR I'm about to send for review: [paste diff].
Review it like a strict senior engineer who's seen this codebase for years. Flag:
- Things that will get nitpicked in code review
- Potential bugs or race conditions I missed
- Tests I should add before submitting
- One simplification that would make the PR shorter without losing functionality
Try in playground →
Full track coming Q3 2026: Claude Code deep-dive, Cursor patterns, AI-assisted debugging workflows, code-review prompts, and how to use AI without producing tech debt.
⚖️
AI for Lawyers
Legal work is high-stakes and citation-sensitive. The win for lawyers is using AI for the volume work — first-pass review, redlining, drafting — while keeping every claim verifiable. These prompts emphasize specificity, citation, and hallucination resistance.
Best tools
Claude
Perplexity
Contract review — first pass
Review this contract from the perspective of [party]. [Paste contract].
Identify:
1. Top 5 clauses that disproportionately favor the other side
2. Standard market terms that are missing
3. Any ambiguous or unenforceable language
4. Three specific redlines you'd propose, with replacement language
Do not include legal disclaimers. Assume the reader is an experienced attorney.
Try in playground →
Case research with citations
Find case law from [jurisdiction] in the last [N years] that addresses [issue]. For each case, give me:
- Case name and citation
- One-sentence holding
- Key facts
- Whether it's currently good law
- A direct quote from the opinion if available
If you can't verify a case exists, say so explicitly — do not invent citations.
Try in playground →
Full track coming Q3 2026: The complete lawyer's AI playbook — verified citation workflows, contract redline systems, deposition prep, and the safe-use patterns that protect privilege and avoid hallucinated cases.
🚀
AI for Founders
Founders wear every hat at once. AI is the cheapest possible head-count — the trick is using it without producing slop. These prompts cover the work founders actually do: pitching, customer interviews, hiring, and thinking through hard decisions.
Best tools
Claude
ChatGPT
Perplexity
Pitch deck stress test
Here's my Series A pitch deck (or any pitch). [Paste slides or describe].
Act as a skeptical lead partner at a top-tier VC. After reviewing it, tell me:
1. The single weakest slide and why
2. Three questions you'd ask in the partner meeting that I should have a great answer for
3. The narrative gap — what doesn't add up about my story?
4. One thing I should cut entirely
Don't be polite.
Try in playground →
Customer interview synthesis
Here are transcripts from 8 customer interviews: [paste].
Synthesize:
1. The 3 most common pain points (with frequency)
2. The exact phrases customers used (not paraphrased)
3. The pain point most people described but didn't ask us to solve
4. Three potential product directions, ranked by what's said most vs what people actually pay for
5. One thing I should ask in the next 5 interviews to verify
Try in playground →
Full track coming Q3 2026: Pitch refinement, fundraising email templates, customer-interview synthesis workflows, hiring rubrics, and a complete bootstrapping operations playbook.
👩🏫
AI for Teachers
Teachers carry an unsustainable workload — lesson planning, differentiation, grading, parent comms. AI well-applied can give you Sundays back. These prompts focus on outputs that actually fit how a classroom runs.
Best tools
Claude
ChatGPT
Differentiated lesson plan
I'm teaching [topic] to [grade level]. Create a 45-minute lesson plan with:
- Hook (5 min)
- Direct instruction (10 min)
- Guided practice (15 min)
- Independent practice (10 min)
- Exit ticket (5 min)
Then differentiate for: (a) a student reading 2 grades below level, (b) a student who finishes early and needs extension, (c) an English-language learner. Keep the differentiated versions on the same topic.
Try in playground →
Parent email — difficult conversation
I need to email a parent about [situation]. The student is [brief context].
Write a professional, warm email that:
- Doesn't bury the lead
- States the specific concern factually
- Names one positive observation about the student
- Proposes a specific next step (call, meeting, plan)
- Ends with collaboration, not blame
Under 200 words. Tone: a caring teacher, not an HR memo.
Try in playground →
Full track coming Q3 2026: A full teacher's AI curriculum — lesson planning workflows, IEP support patterns, rubric design, grading assistance, and the parent-communication library.
📊
AI for Finance
Finance is precision work — but a huge amount of it is repetitive variance analysis, narrative memos, and Excel formula construction. The pattern here is using AI for the prose, the explanation, and the formula scaffolding, while you still own the numbers.
Best tools
Copilot
Claude
Gemini
Variance memo from the numbers
Here's our Q[N] vs plan variance: [paste table — line items, plan, actual, variance, % variance].
Write a 1-page CFO variance memo:
- 3-sentence summary of the quarter
- Top 3 over-plan drivers with hypothesized cause
- Top 3 under-plan drivers with hypothesized cause and recommended action
- One callout for the board that's easy to miss
Tone: confident, concise, data-driven. No corporate hedging.
Try in playground →
Excel formula architect
I have a spreadsheet with [describe columns and rows]. I want to [describe outcome].
Tell me:
1. The exact Excel formula(s) I need, with cell references that match my structure
2. Why this formula is the right choice vs alternatives (XLOOKUP vs INDEX/MATCH, SUMIFS vs SUMPRODUCT, etc.)
3. The one edge case where this formula will silently break, and how to handle it
Don't suggest VBA unless it's truly necessary.
Try in playground →
Full track coming Q3 2026: The complete finance curriculum — Excel + Copilot deep-dive, financial-modeling prompts, audit-ready documentation patterns, and FP&A narrative templates.
🎨
AI for Designers
Designers use AI two ways: for the writing around the work (briefs, presentations, copy), and for the visual exploration itself (image generation, ideation, reference). These prompts cover both — and they keep the human as the taste-maker.
Best tools
Claude
ChatGPT
Creative brief from a vague request
A stakeholder said: "We need a refresh of [thing]. Make it feel more modern."
Turn that into an actual creative brief:
- 3 clarifying questions I should ask them before starting
- The likely actual problem they're trying to solve (not what they said)
- Audience, mood, references, must-avoids
- Definition of done — how will we know it's "more modern"?
- One question they probably haven't thought about
Try in playground →
Image-gen prompt engineering
I want to generate a [type of image] for [use case]. The mood is [adjectives].
Write me 5 image-gen prompts, each progressively more specific:
1. Loose / exploratory
2. Composition-focused
3. Mood + lighting heavy
4. Strong specific reference style
5. Production-ready (lighting, lens, materials, color palette)
For each, tell me what kind of result I'll get and which AI image tool is best for that style.
Try in playground →
Full track coming Q3 2026: Image-generation deep dive (Midjourney, DALL-E, Stable Diffusion), brief-to-presentation pipelines, UX-writing patterns, and the design-research synthesis workflow.
🏗️
AI for Contractors
Contractors lose hours every week to bids, change orders, and writing the same customer email over and over. AI doesn't swing a hammer — but it can turn a 90-minute estimate into a 15-minute one, draft the change-order email while you're still in the truck, and read the specs your client emailed you so you don't have to.
Best tools
Claude
ChatGPT
Copilot
Bid estimate from a walkthrough
I just walked a job. Here are my notes from the walkthrough: [paste].
Help me draft a bid estimate:
1. Itemized line items (labor + materials, separated)
2. Realistic hours per task based on what I described
3. Suggested markup range for this kind of work
4. Three things in my notes that are ambiguous and I should clarify with the client before quoting
5. One risk in this scope I should add a contingency for
Try in playground →
Change order email to the homeowner
The homeowner asked for [change]. Original scope was [original]. The new work adds [hours / materials / cost].
Write a short, professional change-order email:
- Restate what changed and why
- New cost broken out (not just a total)
- New timeline impact (be specific — days, not "a bit longer")
- Ask for written confirmation before we proceed
- Friendly but firm — no "circling back" language
Under 200 words.
Try in playground →
Punch list from job-site photos
[Attach 4-8 photos of the worksite at end-of-day or pre-walkthrough.]
Walk through the photos and build me a punch list:
1. Items that need finishing or correction, by location/room
2. Items that look incomplete but might already be done (flag for me to verify)
3. Any safety issues I should fix before the next inspection
4. Materials I'll need to order to close out the punch list
Be specific. "Touch up paint" is not specific. "Touch up paint above the south kitchen window where the cut is uneven" is.
Try in playground →
Read the architect's spec sheet for me
[Paste or attach the spec sheet / architect's drawings.]
Summarize the spec for me like a foreman would brief a crew:
1. What's the job in plain English — three sentences
2. Materials called out (brand + model where specified)
3. Anything that's atypical or could surprise us on day one
4. Code requirements or inspections referenced
5. Three questions I should ask the architect or GC before we break ground
Try in playground →
Full track coming Q3 2026: The contractor's AI playbook — bid generation, change-order systems, spec-sheet decoding, photo-based punch lists, and customer-comms templates that actually sound like a tradesperson, not a marketing agency.
🔧
AI for Skilled Trades
HVAC, plumbing, electrical, appliance repair. The work isn't done at a desk — but the paperwork around it is endless. AI is fastest at the stuff between the work: diagnostic narrowing from a customer description, code lookups, and writing the quote you'd put off for two days.
Best tools
Claude
ChatGPT
Perplexity
Diagnose from a photo + customer description
[Attach photo of the unit, panel, fixture, or symptom.]
Customer says: "[paste exact words they used]"
Make / model: [if known]
Age: [if known]
Other context: [what they tried, when it started, weather, etc.]
What are the 3 most likely root causes in order of probability? For each:
- What I should check first to confirm
- Estimated repair difficulty (DIY-able, journeyman, master-level)
- Approximate parts cost range
If the photo shows anything unsafe (exposed wires, gas smell hazard, water near electrical), call it out first.
Try in playground →
Code lookup with citation
I'm doing [type of work] in [city/state]. The job involves [specific situation].
Tell me the applicable code requirements — National Electrical Code, International Plumbing Code, International Residential Code, or local amendment if it applies in [jurisdiction].
For each requirement: cite the code section, paraphrase the requirement in plain English, and note if there are common variations or exceptions an inspector might flag.
If you can't verify a specific local amendment, say so — don't guess at code.
Try in playground →
Quote a service call for the customer
The job: [describe what we did].
Time on site: [hours]
Parts used: [list]
Diagnostic time: [hours]
Write a customer-facing quote/invoice that:
- Explains the problem and the fix in plain English (no acronyms)
- Itemizes labor, diagnostic, parts separately
- Includes my standard warranty language
- Sounds like a small business, not a corporate template
- Ends with one specific thing the customer can do to avoid this issue again
Try in playground →
Explain the repair to a customer who's overwhelmed
Here's what I just told a customer needs to be done: [paste technical explanation].
Rewrite it for someone who doesn't know what a [technical term] is. Keep it accurate. Don't talk down to them. Use a real-world analogy where it helps. End with the two numbers that matter most to them: what it'll cost, and how long it'll take.
Try in playground →
Full track coming Q3 2026: The skilled-trades AI playbook — diagnostic workflows with photos, NEC / IPC / IRC code lookups, customer-quote systems, warranty letter templates, and the small-business-owner's invoicing automation.
🚛
AI for Trucking
Owner-operators and small fleets spend half their lives on the phone with dispatch, brokers, shippers, and DOT. AI can take the friction out of the comms — and answer the regulation questions you used to call your buddy about.
Best tools
Claude
Perplexity
ChatGPT
DOT / ELD / HOS question (with citation)
I have a question about [hours-of-service / ELD / weight limits / inspection / something else]: [describe situation].
Answer based on current FMCSA regulations. Cite the specific section of 49 CFR. Tell me:
- The rule in plain English
- How it applies to my specific situation
- Common exceptions I might qualify for
- What an inspector or DOT officer would look for
If the regulation has changed recently, flag the change. If you're not certain a specific interpretation is current, say so — don't guess on regs.
Try in playground →
ETA update / delay message to dispatch or broker
Situation: [describe — weather, traffic, breakdown, mandatory rest, etc.]
Original ETA: [time]
New realistic ETA: [time]
Load / PO #: [number]
Write a short, professional message to dispatch / the broker:
- State the new ETA up front
- One-sentence reason (no excuses, just facts)
- What I'm doing to mitigate (if anything)
- When I'll provide the next update
Keep it under 60 words. Tone: pro driver, not over-apologetic.
Try in playground →
Load evaluation — should I take this run?
Here's the load: [origin → destination, miles, weight, commodity, pay, deadhead]
My truck: [type, current location, available HOS]
Fuel price avg: [if known]
Tell me:
1. Effective rate per mile (loaded + deadhead)
2. How it compares to typical lanes in that region
3. Realistic time on the load with HOS
4. One thing I should ask the broker before accepting
5. Should I take it — and why
Try in playground →
Maintenance log → action list
Here's my recent maintenance log: [paste].
Current odometer: [miles]
Tell me:
1. What's due now based on standard service intervals
2. What's due in the next 30 days
3. Anything that's overdue and I should not put off
4. Patterns I should watch (recurring issues, things wearing earlier than expected)
5. One question I should ask the next time it's in the shop
Try in playground →
Full track coming Q3 2026: The trucking AI playbook — FMCSA/DOT regulation lookups, broker negotiation patterns, rate-per-mile load evaluation, maintenance scheduling, and owner-operator business operations.
🚗
AI for Auto Repair
Independent shops live or die on three things: accurate diagnostics, fast parts sourcing, and explaining the repair to a skeptical customer without losing the sale. AI doesn't replace the lift — but it shortens the time between "car drives in" and "estimate approved."
Best tools
Claude
ChatGPT
Perplexity
OBD-II code → likely causes
Vehicle: [year/make/model/engine/mileage]
DTC code(s): [list]
Symptoms the customer described: [paste]
What I've already verified: [list]
Walk me through:
1. What this code actually means (not just the generic description)
2. Top 3 root causes for this vehicle, in order of probability
3. What I should test/check to narrow it down, in order from fastest to slowest
4. Known issues for this engine where this code appears (TSBs, common patterns)
5. Approximate parts and labor cost range to fix the most likely cause
Try in playground →
Repair estimate for the customer
Vehicle: [year/make/model]
Diagnosed issue: [what's wrong]
Required parts: [list, with OEM vs aftermarket options]
Labor hours per my manual: [X]
Shop rate: [$/hr]
Write a customer-facing repair estimate:
- Plain-English explanation of what's wrong and why it matters
- Itemized parts (OEM vs aftermarket pricing both shown)
- Labor broken out
- Warranty terms
- One related issue I noticed they should know about (but isn't urgent)
- Three-sentence summary of "what happens if you don't fix this"
Tone: trustworthy mechanic, not a high-pressure sales pitch.
Try in playground →
Parts cross-reference + sourcing
I need: [part description, including OEM part number if I have it]
Vehicle: [year/make/model/engine]
Urgency: [in-shop today / customer can wait 2-3 days / etc.]
Help me find:
1. Equivalent aftermarket part numbers (Moog, Bosch, Denso, etc.)
2. Known quality difference (which aftermarket brands are equivalent to OEM, which to avoid)
3. Likely price range for OEM vs aftermarket
4. Suppliers most likely to have it in stock today
5. Compatible parts that fit but aren't an exact match — and the risk if I use them
Try in playground →
"Why should I fix this?" — customer pushback
The customer said: "[paste exactly what they said]"
The repair I recommended: [describe]
The risk of not fixing it: [what happens]
Write me a response that:
- Doesn't restate what I already told them
- Acknowledges their concern (cost, timing, whatever they said) honestly
- Explains the specific risk they're taking by deferring, with a real-world consequence (not a scare tactic)
- Offers one realistic compromise option if there is one (temporary fix, do-the-critical-part-now, etc.)
- Ends with their choice clearly in their hands — no pressure
Try in playground →
Full track coming Q3 2026: The auto repair AI playbook — DTC interpretation libraries, parts-sourcing workflows, customer-estimate templates, objection handling, and shop-management operations.