Curated AI prompts for business, creative, and technical use cases.
Works with any high-quality image generation model, including Nano Banana Pro, Midjourney, DALL-E 3, and Stable Diffusion/FLUX variants. The prompt structure (subject, setting, lighting, camera, mood) is deliberately model-agnostic since it mirrors how professional photographers already brief a shoot.creativeA solo founder running a small ceramics brand needs a hero product image for a new mug listing on Instagram and their Shopify store this week, but has no budget for a physical photoshoot and only a rough phone photo of the prototype to work from.
A structured image-generation prompt that replaces vague requests like 'nice photo of my product' with the exact lighting, setting, and camera detail that separates professional-looking AI photography from obviously synthetic output.
image-generationprompt-engineering
Works best with Claude Opus 5 or GPT-6-class models for reasoning through execution plans and index tradeoffs; any capable coding model can follow the structure, but weaker models tend to suggest indexes without checking whether a query rewrite would be simpler or lower-risk.Developer ToolsA backend engineer notices a dashboard endpoint that used to load in under a second now takes over 4 seconds after the orders table crossed 50 million rows. They have the slow query and an EXPLAIN ANALYZE output in hand, but aren't sure whether the fix is a new index, a query rewrite, or a schema change — and don't want to guess on a production database.
A structured prompt that forces a real root-cause diagnosis of a slow SQL query before recommending any fix — distinguishing safe query rewrites from risky schema migrations.
prompt-engineeringpostgresql
Works well with Claude (Opus or Sonnet) and GPT-6-class models for holding platform-specific tone distinctions; any frontier model handles the structure, but weaker models tend to under-vary hooks without the explicit labeling requirement.marketingA content marketer just published a 1,500-word blog post on why remote-work productivity metrics are broken, and has 30 minutes before their social scheduling tool's weekly queue locks. They need native, non-duplicate posts for LinkedIn, X, and an Instagram caption that each stand alone and drive clicks back to the full piece, without just pasting the same paragraph into three different boxes.
A structured prompt that turns one article into distinct, platform-appropriate social posts — each with a different hook and a labeled source, instead of the same paragraph pasted three times.
prompt-engineeringsocial-media
Works well with Claude (Opus or Sonnet) and GPT-6-class models for holding technical nuance while simplifying; verify domain-specific claims (security, compliance, legal) against the source before sharing externally.Writing & CommunicationA backend engineer just shipped documentation for a new authentication API covering JWT rotation and idempotent retries. The sales team needs to explain these safeguards to a healthcare client asking pointed questions about session security and duplicate-charge protection during a live demo call in two hours — and they can't parse terms like “idempotent” or “token rotation” fast enough to answer confidently.
A structured prompt that translates technical documentation into plain English while flagging exactly which simplified details are safe to lose and which ones could cause a costly mistake.
prompt-engineeringdocumentation
Works best with Claude (Opus or Sonnet) for nuanced copy critique and distinguishing real friction from stylistic preference; GPT-5-class models and Gemini 2.5 Pro also handle this well.marketingA marketing manager is about to increase paid search spend by $10,000/month for a SaaS landing page currently converting at 1.2% against a 3% category benchmark, and needs to know exactly what to fix before sending more traffic to a leaky page.
A structured prompt that audits landing page copy for conversion friction — flagging vague language, missing trust signals, and unclear value propositions — then ranks fixes by impact versus effort.
prompt-engineeringmarketing
Works best with Claude (Opus or Sonnet) for precise reasoning about database internals and lock behavior; GPT-5-class models handle it well too, but verify lock-type claims against your specific database version's documentation before running anything in production.Developer ToolsYour team is two days from a major product launch and needs to add a required boolean column to your 50-million-row users table to support a new feature — but the last time someone ran a similar migration, it locked the table for six minutes and paged the entire on-call rotation during peak traffic.
A structured prompt that analyzes a proposed schema change for locking behavior, backward compatibility, and rollback safety — then produces a step-by-step migration plan that won't take your production database down.
prompt-engineeringdeveloper tools
Works well with GPT-5.4, Claude Opus 5, or Gemini 3 Pro — needs strong world knowledge of how technical fields build on each other, so more capable models produce more accurate prerequisite chains than smaller ones.learningYou're three chapters into a machine learning textbook and hit a section on backpropagation that assumes you're comfortable with the chain rule and Jacobian matrices — but it's been eight years since your last calculus class and you're not sure exactly what you're missing or where to start filling the gap.
A four-part diagnostic prompt that pinpoints the specific prerequisite gaps causing you to get stuck on a hard topic — ordered by dependency, each with a one-line self-test and a fastest-path shortcut.
prompt-engineeringlearning
Works well with GPT-5.4, Claude Opus 5, or Gemini 3 Pro — this task is mostly judgment and tone calibration rather than raw technical capability, so any frontier model handles it well.CareerYou're a mid-level employee who was just told in a short meeting that your role is being eliminated, and HR has handed you a severance agreement with a tight deadline to sign — and you have no idea which terms are fixed and which are actually negotiable.
A five-part negotiation brief that turns a severance agreement into a structured strategy — leverage assessment, prioritized asks, a usable script, and red flags to check before you sign.
CareerNegotiation
Works best with Claude (Sonnet or Opus) for consistent thematic grouping across large volumes of open-text data; GPT-5-class models handle it well too, though tend to over-fragment themes without the explicit six-theme cap.Data AnalysisYou just closed a customer satisfaction survey with 340 open-text responses to "What would make you more likely to renew?" and a board meeting in two days where you need to present findings, not raw data.
A structured prompt that turns hundreds of open-text survey responses into a ranked list of themes with sentiment labels, real customer quotes, and prioritized fixes — not just a wall of raw comments.
ai-productivitysurvey analysis
Works best with Claude (Opus or Sonnet) for nuanced prioritization judgment across ambiguous emails; GPT-5-class models and Gemini 2.5 Pro handle it well too, given a large enough context window to paste the full inbox dump.productivityYou return from a week-long conference to 340 unread emails and a 9am meeting in two hours. You need to know which handful of emails actually require action before that meeting — not spend an hour reading all 340 in order.
A structured prompt that sorts a flooded inbox into what needs a reply today, what can wait, what to delegate, and what to ignore — plus draft replies for the urgent ones.
prompt-engineeringai-productivity
Works well with any current-generation reasoning-capable model (Claude Sonnet/Opus, GPT-6, Gemini 3) — the value is in realistic pushback prediction and situation-specific phrasing, not model-specific capability.A manager needs to address a high-performing but sarcastic team member whose comments have visibly shut down a colleague, ahead of a 1:1 tomorrow, without triggering defensiveness that derails the conversation or escalating to a formal HR complaint.
A prompt that takes a sensitive workplace situation and produces an opening line that won't trigger defensiveness, anticipated pushback with responses for each point, and a concrete next step — not generic conflict-resolution advice.
prompt-engineeringmanagement
Works well with any current-generation coding-capable model (Claude Sonnet/Opus, GPT-6, Gemini 3) — the value is in the structured risk/effort framework, not model-specific code generation ability.A senior engineer inherits a payment webhook handler that has caused three production incidents in the last quarter and needs to convince a skeptical team lead to allocate cleanup time instead of just adding another patch.
A prompt that takes code you know has problems and produces a risk-versus-effort breakdown, concrete before/after examples, and a phased plan — instead of a vague “we should rewrite this”.