Trending AI Prompts for Creative Content & Ideas
Most popular, high-performing AI prompts this week based on creator and developer engagement.
Keep your AI agents reliable, accountable, and affordable with owners, approval gates, monitoring, and a kill switch.
Recommended Tech Stack
A frontier model such as Claude Opus 5.5 or GPT-6 Sol to draft the runbook, an agent or workflow tool you already use (LangGraph, CrewAI, or n8n), MCP connections with least-privilege access, and a logging or tracing tool. Confirm permissions, data handling, and approval rules with your team before any agent goes live.
Quick Implementation Guide
Paste the prompt with your workflow filled in, and ask the AI to state its assumptions and list missing details. Register every agent with an owner, permissions, and risk level, then add approval gates and the kill switch to the riskiest agent first. Run a regression test set after every prompt or model change, and review the success, error, and cost metrics each week.
Real-World Use Case Example
Use this prompt to set up monitoring, owners, and approval rules for a team of AI agents that handle customer emails, research, or reporting. Treat AI output as a draft and verify it against your real permissions, logs, and compliance requirements.
Act as an AI agent operations (AgentOps) lead who keeps production AI agents reliable, accountable, and affordable. Build an operations runbook for the agents used in [business/workflow/tools]. Include an agent registry (name, purpose, owner, tools, permissions, risk level), a named human owner for every agent, clear definitions of success and failure, approval gates for risky actions such as sending messages, spending money, or editing live data, a daily and weekly monitoring routine, logging and trace requirements so every action can be explained later, alert thresholds for loops, cost spikes, and unusual behavior, a kill-switch and rollback procedure, an incident-response template, a regression test set to run after every prompt or model change, a model-upgrade checklist for when new versions launch, monthly metrics (task success rate, escalation rate, cost per task, error rate), and a 30-day setup plan that starts with the riskiest agent.
Recommended Tech Stack: A frontier model such as Claude Opus 5.5 or GPT-6 Sol to draft the runbook, an agent or workflow tool you already use (LangGraph, CrewAI, or n8n), MCP connections with least-privilege access, and a logging or tracing tool. Confirm permissions, data handling, and approval rules with your team before any agent goes live.
Quick Implementation Guide: Paste the prompt with your workflow filled in, and ask the AI to state its assumptions and list missing details. Register every agent with an owner, permissions, and risk level, then add approval gates and the kill switch to the riskiest agent first. Run a regression test set after every prompt or model change, and review the success, error, and cost metrics each week.
Real-World Use Case Example: Use this prompt to set up monitoring, owners, and approval rules for a team of AI agents that handle customer emails, research, or reporting. Treat AI output as a draft and verify it against your real permissions, logs, and compliance requirements.
Act as an AI agent operations (AgentOps) lead who keeps production AI agents reliable, accountable, and affordable. Build an operations runbook for the agents used in [business/workflow/tools]. Include an agent registry (name, purpose, owner, tools, permissions, risk level), a named human owner for every agent, clear definitions of success and failure, approval gates for risky actions such as sending messages, spending money, or editing live data, a daily and weekly monitoring routine, logging and trace requirements so every action can be explained later, alert thresholds for loops, cost spikes, and unusual behavior, a kill-switch and rollback procedure, an incident-response template, a regression test set to run after every prompt or model change, a model-upgrade checklist for when new versions launch, monthly metrics (task success rate, escalation rate, cost per task, error rate), and a 30-day setup plan that starts with the riskiest agent.
Recommended Tech Stack: A frontier model such as Claude Opus 5.5 or GPT-6 Sol to draft the runbook, an agent or workflow tool you already use (LangGraph, CrewAI, or n8n), MCP connections with least-privilege access, and a logging or tracing tool. Confirm permissions, data handling, and approval rules with your team before any agent goes live.
Quick Implementation Guide: Paste the prompt with your workflow filled in, and ask the AI to state its assumptions and list missing details. Register every agent with an owner, permissions, and risk level, then add approval gates and the kill switch to the riskiest agent first. Run a regression test set after every prompt or model change, and review the success, error, and cost metrics each week.
Real-World Use Case Example: Use this prompt to set up monitoring, owners, and approval rules for a team of AI agents that handle customer emails, research, or reporting. Treat AI output as a draft and verify it against your real permissions, logs, and compliance requirements.
Protect your team, family, or channel from fake video calls, cloned voices, and AI-written phishing.
Recommended Tech Stack
Any strong reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) to draft the playbook, plus your existing email security, multi-factor authentication, and password manager. Confirm payment approval rules and reporting channels with your bank, IT team, or local cybercrime authorities.
Quick Implementation Guide
Fill in who you are protecting and which workflows are risky, such as payments, password resets, and urgent requests from executives. Set up call-back verification and a shared code word, then run a short practice drill with your team or family. Review the playbook every month and update it after any suspicious message, call, or invoice.
Real-World Use Case Example
Use this prompt to protect a small business or family from fake CEO voice calls, cloned voices, and fraudulent invoices. Treat AI output as a starting point and test every verification step with real people before relying on it.
Act as a fraud-prevention and security-awareness specialist focused on AI-enabled impersonation. Build a defense playbook for [company/family/creator account] against deepfake video calls, cloned voices, fake invoices, and AI-written phishing. Include the most likely attack scenarios for my situation in plain language, warning signs to spot, a verification protocol for urgent requests (call back on a known number, shared code words, second-person approval for payments), rules for payment, password-reset, and account-recovery requests, account and email hardening steps, how to limit public voice and video samples that could be misused, a short staff or family training script with practice drills, an incident-response checklist for the first hour after a suspected scam, reporting and evidence-preservation steps, and a monthly review routine. Focus only on defense and awareness. Do not create deceptive content or imitate any real person.
Recommended Tech Stack: Any strong reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) to draft the playbook, plus your existing email security, multi-factor authentication, and password manager. Confirm payment approval rules and reporting channels with your bank, IT team, or local cybercrime authorities.
Quick Implementation Guide: Fill in who you are protecting and which workflows are risky, such as payments, password resets, and urgent requests from executives. Set up call-back verification and a shared code word, then run a short practice drill with your team or family. Review the playbook every month and update it after any suspicious message, call, or invoice.
Real-World Use Case Example: Use this prompt to protect a small business or family from fake CEO voice calls, cloned voices, and fraudulent invoices. Treat AI output as a starting point and test every verification step with real people before relying on it.
Act as a fraud-prevention and security-awareness specialist focused on AI-enabled impersonation. Build a defense playbook for [company/family/creator account] against deepfake video calls, cloned voices, fake invoices, and AI-written phishing. Include the most likely attack scenarios for my situation in plain language, warning signs to spot, a verification protocol for urgent requests (call back on a known number, shared code words, second-person approval for payments), rules for payment, password-reset, and account-recovery requests, account and email hardening steps, how to limit public voice and video samples that could be misused, a short staff or family training script with practice drills, an incident-response checklist for the first hour after a suspected scam, reporting and evidence-preservation steps, and a monthly review routine. Focus only on defense and awareness. Do not create deceptive content or imitate any real person.
Recommended Tech Stack: Any strong reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) to draft the playbook, plus your existing email security, multi-factor authentication, and password manager. Confirm payment approval rules and reporting channels with your bank, IT team, or local cybercrime authorities.
Quick Implementation Guide: Fill in who you are protecting and which workflows are risky, such as payments, password resets, and urgent requests from executives. Set up call-back verification and a shared code word, then run a short practice drill with your team or family. Review the playbook every month and update it after any suspicious message, call, or invoice.
Real-World Use Case Example: Use this prompt to protect a small business or family from fake CEO voice calls, cloned voices, and fraudulent invoices. Treat AI output as a starting point and test every verification step with real people before relying on it.
Get your site cited by AI answer engines and readable by AI agents that research and buy for users.
Recommended Tech Stack
ChatGPT, Gemini, Claude, and Perplexity to check how your brand appears in AI answers, Google Search Console and a site crawler for technical checks, and Schema.org structured data in JSON-LD. Verify every change against current search-engine guidelines before publishing.
Quick Implementation Guide
Run the prompt with your URL and top competitors, and ask the AI to state assumptions and list missing information. Test your 30 buyer-style questions in each AI tool and record which sites get cited and why. Add direct-answer blocks, FAQ sections, and schema to your best pages, then track AI referrals and brand mentions every month.
Real-World Use Case Example
Use this prompt to audit a store, blog, or local service site so AI answer engines can cite it and AI shopping agents can read its prices and policies. Treat AI output as a starting point and verify rankings and traffic changes with real analytics.
Act as a generative engine optimization (GEO) and technical SEO specialist who prepares websites for AI answer engines and for AI agents that research, compare, and buy on behalf of users. Audit [URL/niche/store] and produce a complete plan. Include how ChatGPT, Gemini, Claude, Perplexity, and Google AI Overviews currently describe my brand, which pages and products AI systems can and cannot parse, direct-answer paragraphs, comparison tables, and FAQ blocks to add, structured data (Product, FAQ, Organization, Review, Article) to implement, clear pricing, availability, shipping, and return details that agents can read, author and brand trust signals, 30 buyer-style questions my pages should answer, a robots.txt and AI-crawler access policy, citation and mention-building tactics, a tracking plan for AI referrals and brand mentions, mistakes to avoid such as thin auto-generated pages, and a 90-day roadmap ranked by impact and effort.
Recommended Tech Stack: ChatGPT, Gemini, Claude, and Perplexity to check how your brand appears in AI answers, Google Search Console and a site crawler for technical checks, and Schema.org structured data in JSON-LD. Verify every change against current search-engine guidelines before publishing.
Quick Implementation Guide: Run the prompt with your URL and top competitors, and ask the AI to state assumptions and list missing information. Test your 30 buyer-style questions in each AI tool and record which sites get cited and why. Add direct-answer blocks, FAQ sections, and schema to your best pages, then track AI referrals and brand mentions every month.
Real-World Use Case Example: Use this prompt to audit a store, blog, or local service site so AI answer engines can cite it and AI shopping agents can read its prices and policies. Treat AI output as a starting point and verify rankings and traffic changes with real analytics.
Act as a generative engine optimization (GEO) and technical SEO specialist who prepares websites for AI answer engines and for AI agents that research, compare, and buy on behalf of users. Audit [URL/niche/store] and produce a complete plan. Include how ChatGPT, Gemini, Claude, Perplexity, and Google AI Overviews currently describe my brand, which pages and products AI systems can and cannot parse, direct-answer paragraphs, comparison tables, and FAQ blocks to add, structured data (Product, FAQ, Organization, Review, Article) to implement, clear pricing, availability, shipping, and return details that agents can read, author and brand trust signals, 30 buyer-style questions my pages should answer, a robots.txt and AI-crawler access policy, citation and mention-building tactics, a tracking plan for AI referrals and brand mentions, mistakes to avoid such as thin auto-generated pages, and a 90-day roadmap ranked by impact and effort.
Recommended Tech Stack: ChatGPT, Gemini, Claude, and Perplexity to check how your brand appears in AI answers, Google Search Console and a site crawler for technical checks, and Schema.org structured data in JSON-LD. Verify every change against current search-engine guidelines before publishing.
Quick Implementation Guide: Run the prompt with your URL and top competitors, and ask the AI to state assumptions and list missing information. Test your 30 buyer-style questions in each AI tool and record which sites get cited and why. Add direct-answer blocks, FAQ sections, and schema to your best pages, then track AI referrals and brand mentions every month.
Real-World Use Case Example: Use this prompt to audit a store, blog, or local service site so AI answer engines can cite it and AI shopping agents can read its prices and policies. Treat AI output as a starting point and verify rankings and traffic changes with real analytics.
Write the spec first, let coding agents build in parallel, and catch bugs with automated reviews before you merge.
Recommended Tech Stack
Claude Code, Codex, or Cursor with a frontier model (Claude Opus 5.5 or GPT-6 Sol), Git with branch protection, an automated test runner with CI, and an AI pull-request review step. Confirm secrets handling and repository permissions before giving any agent access.
Quick Implementation Guide
Paste the prompt with your stack and project details to generate a spec template and a project instructions file. Write a spec for one small feature, let the agent propose a plan, then build it on its own branch with tests first. Review each pull request with the checklist, and update the instructions file every week with the agent's repeated mistakes.
Real-World Use Case Example
Use this prompt to build a new feature in a web app with coding agents while keeping every change small, tested, and reviewed. Treat AI output as a starting point and verify functionality, security, and performance before merging.
Act as a staff engineer who ships software with agentic coding tools such as Claude Code, Codex, and Cursor. Design a spec-driven development workflow for my project [describe stack and project]. Include a spec template (goal, user stories, constraints, acceptance tests) that the agent must follow, a project instructions file with architecture, naming, folder rules, and do and don't lists, how to split work into small parallel tasks on separate branches, a plan-first step where the agent proposes changes before editing any file, test-first prompts, an automated pull-request review checklist covering bugs, security, performance, and test coverage, rules for when the agent must stop and ask a human, secrets and permission boundaries, how to prevent scope creep and unrequested refactors, a debugging prompt for failed builds, metrics for time saved and defects caught, and a weekly routine that turns repeated agent mistakes into new rules.
Recommended Tech Stack: Claude Code, Codex, or Cursor with a frontier model (Claude Opus 5.5 or GPT-6 Sol), Git with branch protection, an automated test runner with CI, and an AI pull-request review step. Confirm secrets handling and repository permissions before giving any agent access.
Quick Implementation Guide: Paste the prompt with your stack and project details to generate a spec template and a project instructions file. Write a spec for one small feature, let the agent propose a plan, then build it on its own branch with tests first. Review each pull request with the checklist, and update the instructions file every week with the agent's repeated mistakes.
Real-World Use Case Example: Use this prompt to build a new feature in a web app with coding agents while keeping every change small, tested, and reviewed. Treat AI output as a starting point and verify functionality, security, and performance before merging.
Act as a staff engineer who ships software with agentic coding tools such as Claude Code, Codex, and Cursor. Design a spec-driven development workflow for my project [describe stack and project]. Include a spec template (goal, user stories, constraints, acceptance tests) that the agent must follow, a project instructions file with architecture, naming, folder rules, and do and don't lists, how to split work into small parallel tasks on separate branches, a plan-first step where the agent proposes changes before editing any file, test-first prompts, an automated pull-request review checklist covering bugs, security, performance, and test coverage, rules for when the agent must stop and ask a human, secrets and permission boundaries, how to prevent scope creep and unrequested refactors, a debugging prompt for failed builds, metrics for time saved and defects caught, and a weekly routine that turns repeated agent mistakes into new rules.
Recommended Tech Stack: Claude Code, Codex, or Cursor with a frontier model (Claude Opus 5.5 or GPT-6 Sol), Git with branch protection, an automated test runner with CI, and an AI pull-request review step. Confirm secrets handling and repository permissions before giving any agent access.
Quick Implementation Guide: Paste the prompt with your stack and project details to generate a spec template and a project instructions file. Write a spec for one small feature, let the agent propose a plan, then build it on its own branch with tests first. Review each pull request with the checklist, and update the instructions file every week with the agent's repeated mistakes.
Real-World Use Case Example: Use this prompt to build a new feature in a web app with coding agents while keeping every change small, tested, and reviewed. Treat AI output as a starting point and verify functionality, security, and performance before merging.
Turn up to three product photos into a realistic 9:16 AI commercial that keeps the logo, colors, and shape accurate.
Recommended Tech Stack
Veo in the Gemini app or Flow (reference images, up to three per video, 8-second clips), a video editor such as CapCut or VN for titles and for joining clips, and clean product photos on plain backgrounds. Check your tool's current limits, pricing, and content rules before generating.
Quick Implementation Guide
Upload up to three clear images of the same product and fill in the bracketed details in the prompt. Generate the clip, check the logo, text, and shape against your images, and regenerate with the negative prompt if anything is distorted. Chain the 8-second clips in your editor, then add the title text, music, and captions.
Real-World Use Case Example
Use this prompt to make a realistic 9:16 product video for headphones, shoes, or gadgets from your existing product photos. Treat AI output as a starting point and check every frame for accuracy before publishing.
Act as an AI video director who specializes in realistic product commercials for image-to-video and reference-image tools such as Veo in Gemini. I will upload up to three reference images of one product [product name, color, and what each image shows]. Write a production-ready prompt for a vertical 9:16, 8-second clip that keeps the product's shape, colors, text, and logo identical to the references. Include a shot-by-shot timeline with timestamps, camera movement (orbit, push-in, pull-back), lens feel and shallow depth of field, studio lighting and surface, product action (rotate, unfold, reveal), audio direction (music mood and subtle sound effects), which reference image to use for each shot, a negative prompt to avoid warped logos, garbled text, extra parts, flicker, hands, and cables, a plan for chaining three 8-second clips into one 24-second video, and two alternative versions to test. Keep title text out of the generated video and add it later in a video editor.
Recommended Tech Stack: Veo in the Gemini app or Flow (reference images, up to three per video, 8-second clips), a video editor such as CapCut or VN for titles and for joining clips, and clean product photos on plain backgrounds. Check your tool's current limits, pricing, and content rules before generating.
Quick Implementation Guide: Upload up to three clear images of the same product and fill in the bracketed details in the prompt. Generate the clip, check the logo, text, and shape against your images, and regenerate with the negative prompt if anything is distorted. Chain the 8-second clips in your editor, then add the title text, music, and captions.
Real-World Use Case Example: Use this prompt to make a realistic 9:16 product video for headphones, shoes, or gadgets from your existing product photos. Treat AI output as a starting point and check every frame for accuracy before publishing.
Act as an AI video director who specializes in realistic product commercials for image-to-video and reference-image tools such as Veo in Gemini. I will upload up to three reference images of one product [product name, color, and what each image shows]. Write a production-ready prompt for a vertical 9:16, 8-second clip that keeps the product's shape, colors, text, and logo identical to the references. Include a shot-by-shot timeline with timestamps, camera movement (orbit, push-in, pull-back), lens feel and shallow depth of field, studio lighting and surface, product action (rotate, unfold, reveal), audio direction (music mood and subtle sound effects), which reference image to use for each shot, a negative prompt to avoid warped logos, garbled text, extra parts, flicker, hands, and cables, a plan for chaining three 8-second clips into one 24-second video, and two alternative versions to test. Keep title text out of the generated video and add it later in a video editor.
Recommended Tech Stack: Veo in the Gemini app or Flow (reference images, up to three per video, 8-second clips), a video editor such as CapCut or VN for titles and for joining clips, and clean product photos on plain backgrounds. Check your tool's current limits, pricing, and content rules before generating.
Quick Implementation Guide: Upload up to three clear images of the same product and fill in the bracketed details in the prompt. Generate the clip, check the logo, text, and shape against your images, and regenerate with the negative prompt if anything is distorted. Chain the 8-second clips in your editor, then add the title text, music, and captions.
Real-World Use Case Example: Use this prompt to make a realistic 9:16 product video for headphones, shoes, or gadgets from your existing product photos. Treat AI output as a starting point and check every frame for accuracy before publishing.
Run private open-weight models on your own computer and know when a cloud model is the better choice.
Recommended Tech Stack
Ollama or LM Studio, an open-weight model such as DeepSeek V4, GLM-5.2, or Meta's Muse Glimmer (30B), and a computer with enough RAM or GPU memory for the model size you choose. Download models only from official sources and confirm each model's license.
Quick Implementation Guide
Share your hardware details and use cases, and ask the AI which model sizes realistically fit your computer. Install your chosen model, then test it on 10 of your real tasks and compare the results with a cloud model. Add your own documents with local retrieval (RAG), and review new open-weight releases every month.
Real-World Use Case Example
Use this prompt to run a private assistant on your own computer for writing, coding, or document questions without sending data to a cloud service. Treat AI output as a starting point and verify speed and quality on your real work.
Act as a local-AI engineer who helps people run open-weight models privately on their own hardware. Build a setup guide for my computer [CPU, GPU, RAM] and my use cases [writing, coding, document Q&A, offline assistant]. Include which model sizes realistically fit my hardware and which do not, a comparison of current open-weight options (for example DeepSeek V4, GLM-5.2, and Meta's Muse Glimmer) by size, license, context length, and strengths, step-by-step installation with Ollama or LM Studio, quantization choices and the quality-versus-speed tradeoff, a simple test to compare local output against a cloud model on my real tasks, how to add my own documents safely with local retrieval (RAG), privacy and security checks such as trusted model sources only and no data leaving the machine, when to switch to a cloud model instead, expected speed and electricity cost, and a monthly routine to review new releases.
Recommended Tech Stack: Ollama or LM Studio, an open-weight model such as DeepSeek V4, GLM-5.2, or Meta's Muse Glimmer (30B), and a computer with enough RAM or GPU memory for the model size you choose. Download models only from official sources and confirm each model's license.
Quick Implementation Guide: Share your hardware details and use cases, and ask the AI which model sizes realistically fit your computer. Install your chosen model, then test it on 10 of your real tasks and compare the results with a cloud model. Add your own documents with local retrieval (RAG), and review new open-weight releases every month.
Real-World Use Case Example: Use this prompt to run a private assistant on your own computer for writing, coding, or document questions without sending data to a cloud service. Treat AI output as a starting point and verify speed and quality on your real work.
Act as a local-AI engineer who helps people run open-weight models privately on their own hardware. Build a setup guide for my computer [CPU, GPU, RAM] and my use cases [writing, coding, document Q&A, offline assistant]. Include which model sizes realistically fit my hardware and which do not, a comparison of current open-weight options (for example DeepSeek V4, GLM-5.2, and Meta's Muse Glimmer) by size, license, context length, and strengths, step-by-step installation with Ollama or LM Studio, quantization choices and the quality-versus-speed tradeoff, a simple test to compare local output against a cloud model on my real tasks, how to add my own documents safely with local retrieval (RAG), privacy and security checks such as trusted model sources only and no data leaving the machine, when to switch to a cloud model instead, expected speed and electricity cost, and a monthly routine to review new releases.
Recommended Tech Stack: Ollama or LM Studio, an open-weight model such as DeepSeek V4, GLM-5.2, or Meta's Muse Glimmer (30B), and a computer with enough RAM or GPU memory for the model size you choose. Download models only from official sources and confirm each model's license.
Quick Implementation Guide: Share your hardware details and use cases, and ask the AI which model sizes realistically fit your computer. Install your chosen model, then test it on 10 of your real tasks and compare the results with a cloud model. Add your own documents with local retrieval (RAG), and review new open-weight releases every month.
Real-World Use Case Example: Use this prompt to run a private assistant on your own computer for writing, coding, or document questions without sending data to a cloud service. Treat AI output as a starting point and verify speed and quality on your real work.
Send each task to the right model tier and effort level, and cut your AI bill without losing quality.
Recommended Tech Stack
Frontier models (GPT-6 Sol, Claude Opus 5.5) for hard tasks, lower-cost models (GPT-6 Luna, Claude Sonnet 5.5, Claude Haiku 5.5, Gemini 3.8 Flash) for volume work, DeepSeek V4 for open-weight use, and a simple routing script or gateway. Check current prices, limits, and data policies on each vendor's site before committing.
Quick Implementation Guide
List your tasks with their monthly volume, and ask the AI to score each by difficulty, speed, privacy, and budget. Run the 20-sample test on two or three models and keep the cheapest one that meets your quality bar. Set up routing with fallback and escalation rules, and review prices and new models once a month.
Real-World Use Case Example
Use this prompt to cut a monthly AI bill by sending simple tasks to cheaper models and only hard tasks to frontier models. Treat AI output as a starting point and verify costs with real usage data.
Act as an AI solutions consultant who builds cost-efficient multi-model strategies for small teams. Review my tasks [list: writing, coding, support chat, data extraction, research, image, video, or audio generation] and design a routing plan across current model tiers: frontier models for hard reasoning and agents (for example GPT-6 Sol or Claude Opus 5.5), balanced and low-cost models for high-volume work (for example GPT-6 Luna, Claude Sonnet 5.5, Claude Haiku 5.5, or Gemini 3.8 Flash), and open-weight models for private or very high-volume jobs (for example DeepSeek V4). Include a difficulty, speed, privacy, and budget score for each task, which tier handles it and which thinking or effort level to set, fallback and escalation rules, token-saving methods (prompt caching, shorter context, structured outputs, batching), a 20-sample test to compare models on my real work, estimated monthly cost at low, medium, and high usage, vendor lock-in risks and how to avoid them, and a monthly review routine as prices and models change.
Recommended Tech Stack: Frontier models (GPT-6 Sol, Claude Opus 5.5) for hard tasks, lower-cost models (GPT-6 Luna, Claude Sonnet 5.5, Claude Haiku 5.5, Gemini 3.8 Flash) for volume work, DeepSeek V4 for open-weight use, and a simple routing script or gateway. Check current prices, limits, and data policies on each vendor's site before committing.
Quick Implementation Guide: List your tasks with their monthly volume, and ask the AI to score each by difficulty, speed, privacy, and budget. Run the 20-sample test on two or three models and keep the cheapest one that meets your quality bar. Set up routing with fallback and escalation rules, and review prices and new models once a month.
Real-World Use Case Example: Use this prompt to cut a monthly AI bill by sending simple tasks to cheaper models and only hard tasks to frontier models. Treat AI output as a starting point and verify costs with real usage data.
Act as an AI solutions consultant who builds cost-efficient multi-model strategies for small teams. Review my tasks [list: writing, coding, support chat, data extraction, research, image, video, or audio generation] and design a routing plan across current model tiers: frontier models for hard reasoning and agents (for example GPT-6 Sol or Claude Opus 5.5), balanced and low-cost models for high-volume work (for example GPT-6 Luna, Claude Sonnet 5.5, Claude Haiku 5.5, or Gemini 3.8 Flash), and open-weight models for private or very high-volume jobs (for example DeepSeek V4). Include a difficulty, speed, privacy, and budget score for each task, which tier handles it and which thinking or effort level to set, fallback and escalation rules, token-saving methods (prompt caching, shorter context, structured outputs, batching), a 20-sample test to compare models on my real work, estimated monthly cost at low, medium, and high usage, vendor lock-in risks and how to avoid them, and a monthly review routine as prices and models change.
Recommended Tech Stack: Frontier models (GPT-6 Sol, Claude Opus 5.5) for hard tasks, lower-cost models (GPT-6 Luna, Claude Sonnet 5.5, Claude Haiku 5.5, Gemini 3.8 Flash) for volume work, DeepSeek V4 for open-weight use, and a simple routing script or gateway. Check current prices, limits, and data policies on each vendor's site before committing.
Quick Implementation Guide: List your tasks with their monthly volume, and ask the AI to score each by difficulty, speed, privacy, and budget. Run the 20-sample test on two or three models and keep the cheapest one that meets your quality bar. Set up routing with fallback and escalation rules, and review prices and new models once a month.
Real-World Use Case Example: Use this prompt to cut a monthly AI bill by sending simple tasks to cheaper models and only hard tasks to frontier models. Treat AI output as a starting point and verify costs with real usage data.
Build a reusable sound identity with music prompts, a voice profile, and platform-ready voiceover scripts.
Recommended Tech Stack
Lyria 3.5 for music, Gemini 3.8 Flash TTS or another text-to-speech tool for voiceover, and an editor such as CapCut, Audacity, or DaVinci Resolve for mixing. Use only voices and music you have rights to, and check each platform's rules for disclosing AI-generated audio.
Quick Implementation Guide
Fill in your channel or brand details and generate the sound identity, voice profile, and three music prompts. Create the intro, background bed, and outro, then record or generate the voiceover from the rewritten script. Mix everything in your editor, add the AI-audio disclosure line, and run the quality checklist before publishing.
Real-World Use Case Example
Use this prompt to build a consistent intro, background music, and voice for a YouTube channel, Shorts, or podcast. Treat AI output as a starting point and listen carefully for robotic delivery and pronunciation errors.
Act as an audio producer and prompt engineer for AI music and voice tools (for example Lyria 3.5 and Gemini text-to-speech). Create a reusable audio brand kit for [channel/brand/product]. Include a sound identity (genre, mood, tempo, instruments), three ready-to-paste music prompts for an intro, a background bed, and an outro with timing, a voice profile (age range, tone, accent, pace, energy), my script rewritten for spoken delivery in a YouTube version, a Shorts/Reels version under 45 seconds, and a podcast intro version, pronunciation and pacing notes, Hindi or other language adaptation notes if needed, negative prompts to avoid robotic vocals and muddy mixes, loudness and mixing guidance for each platform, a disclosure line for AI-generated audio, and a pre-publish quality checklist. Use only original lyrics and melodies, and only voices I have the rights to use.
Recommended Tech Stack: Lyria 3.5 for music, Gemini 3.8 Flash TTS or another text-to-speech tool for voiceover, and an editor such as CapCut, Audacity, or DaVinci Resolve for mixing. Use only voices and music you have rights to, and check each platform's rules for disclosing AI-generated audio.
Quick Implementation Guide: Fill in your channel or brand details and generate the sound identity, voice profile, and three music prompts. Create the intro, background bed, and outro, then record or generate the voiceover from the rewritten script. Mix everything in your editor, add the AI-audio disclosure line, and run the quality checklist before publishing.
Real-World Use Case Example: Use this prompt to build a consistent intro, background music, and voice for a YouTube channel, Shorts, or podcast. Treat AI output as a starting point and listen carefully for robotic delivery and pronunciation errors.
Act as an audio producer and prompt engineer for AI music and voice tools (for example Lyria 3.5 and Gemini text-to-speech). Create a reusable audio brand kit for [channel/brand/product]. Include a sound identity (genre, mood, tempo, instruments), three ready-to-paste music prompts for an intro, a background bed, and an outro with timing, a voice profile (age range, tone, accent, pace, energy), my script rewritten for spoken delivery in a YouTube version, a Shorts/Reels version under 45 seconds, and a podcast intro version, pronunciation and pacing notes, Hindi or other language adaptation notes if needed, negative prompts to avoid robotic vocals and muddy mixes, loudness and mixing guidance for each platform, a disclosure line for AI-generated audio, and a pre-publish quality checklist. Use only original lyrics and melodies, and only voices I have the rights to use.
Recommended Tech Stack: Lyria 3.5 for music, Gemini 3.8 Flash TTS or another text-to-speech tool for voiceover, and an editor such as CapCut, Audacity, or DaVinci Resolve for mixing. Use only voices and music you have rights to, and check each platform's rules for disclosing AI-generated audio.
Quick Implementation Guide: Fill in your channel or brand details and generate the sound identity, voice profile, and three music prompts. Create the intro, background bed, and outro, then record or generate the voiceover from the rewritten script. Mix everything in your editor, add the AI-audio disclosure line, and run the quality checklist before publishing.
Real-World Use Case Example: Use this prompt to build a consistent intro, background music, and voice for a YouTube channel, Shorts, or podcast. Treat AI output as a starting point and listen carefully for robotic delivery and pronunciation errors.
Test an AI-powered gadget, wearable, or robot idea before you spend money building it.
Recommended Tech Stack
A reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) for analysis, off-the-shelf boards such as Raspberry Pi or ESP32 for prototyping, a 3D printer for the shell, and an existing AI API. Confirm safety, privacy, and certification requirements for your target markets before selling anything.
Quick Implementation Guide
Describe the device and target user, and ask the AI to state assumptions and list missing information. Check the cost, privacy, and certification risks it finds, then compare your idea with three existing products. Build the low-cost prototype and run the 90-day test with 10 target users before spending on production.
Real-World Use Case Example
Use this prompt to test an idea for a wearable, smart-glasses feature, or home robot before building it. Treat every number as an estimate and verify costs, regulations, and demand with real suppliers and users.
Act as a product strategist and hardware engineer who evaluates AI-powered physical products such as wearables, smart glasses, home robots, and pocket AI devices. Validate my product idea [describe the device and target user]. Include the real problem it solves and who pays for it, why it needs hardware instead of an app, what should run on the device versus in the cloud, sensors and components needed with rough cost ranges, privacy and safety risks (cameras, microphones, children, data storage) and how to design around them, regulatory and certification points to check in my target markets, battery, weight, and heat limits, a comparison with three existing products, a low-cost prototype plan using off-the-shelf boards, a 3D-printed shell, and existing AI APIs, a 90-day test plan with 10 target users, a pricing and margin estimate, and the top five reasons this could fail. Be honest about weaknesses and clearly mark every number as an estimate.
Recommended Tech Stack: A reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) for analysis, off-the-shelf boards such as Raspberry Pi or ESP32 for prototyping, a 3D printer for the shell, and an existing AI API. Confirm safety, privacy, and certification requirements for your target markets before selling anything.
Quick Implementation Guide: Describe the device and target user, and ask the AI to state assumptions and list missing information. Check the cost, privacy, and certification risks it finds, then compare your idea with three existing products. Build the low-cost prototype and run the 90-day test with 10 target users before spending on production.
Real-World Use Case Example: Use this prompt to test an idea for a wearable, smart-glasses feature, or home robot before building it. Treat every number as an estimate and verify costs, regulations, and demand with real suppliers and users.
Act as a product strategist and hardware engineer who evaluates AI-powered physical products such as wearables, smart glasses, home robots, and pocket AI devices. Validate my product idea [describe the device and target user]. Include the real problem it solves and who pays for it, why it needs hardware instead of an app, what should run on the device versus in the cloud, sensors and components needed with rough cost ranges, privacy and safety risks (cameras, microphones, children, data storage) and how to design around them, regulatory and certification points to check in my target markets, battery, weight, and heat limits, a comparison with three existing products, a low-cost prototype plan using off-the-shelf boards, a 3D-printed shell, and existing AI APIs, a 90-day test plan with 10 target users, a pricing and margin estimate, and the top five reasons this could fail. Be honest about weaknesses and clearly mark every number as an estimate.
Recommended Tech Stack: A reasoning model (Claude Opus 5.5, GPT-6 Sol, or Gemini 3.8 Flash) for analysis, off-the-shelf boards such as Raspberry Pi or ESP32 for prototyping, a 3D printer for the shell, and an existing AI API. Confirm safety, privacy, and certification requirements for your target markets before selling anything.
Quick Implementation Guide: Describe the device and target user, and ask the AI to state assumptions and list missing information. Check the cost, privacy, and certification risks it finds, then compare your idea with three existing products. Build the low-cost prototype and run the 90-day test with 10 target users before spending on production.
Real-World Use Case Example: Use this prompt to test an idea for a wearable, smart-glasses feature, or home robot before building it. Treat every number as an estimate and verify costs, regulations, and demand with real suppliers and users.
Design an always-on AI assistant that remembers context and acts across apps without overstepping.
Recommended Tech Stack
An assistant or agent platform with memory controls, powered by a frontier model (Claude Sonnet 5.5, GPT-6 Sol, or Gemini 3.8 Flash), with connected apps (email, calendar, notes) given least-privilege access. Check each tool's memory, privacy, and deletion settings before connecting any account.
Quick Implementation Guide
Describe your role and daily routines, and ask the AI to state assumptions and list missing requirements. Set the memory rules, permissions, and approval list, then connect one app at a time. Run the 30-day trial, review remembered items every week, and delete anything wrong or outdated.
Real-World Use Case Example
Use this prompt to plan an always-on assistant that handles your morning brief, inbox triage, and follow-ups without acting on its own for risky tasks. Treat AI output as a starting point and verify permissions and memory behavior in the real tool.
Act as a designer of persistent personal AI assistants that remember context and act across apps. Design my always-on assistant setup for [role/work/life areas]. Include what it should remember and what it must never store (sensitive personal details and other people's private information), how I can view, edit, and delete its memory, which apps and data it can access with least-privilege permissions, tasks it may do alone versus tasks that need my approval (sending messages, spending money, deleting files), a daily routine (morning brief, inbox triage, task follow-ups), protection against prompt injection from emails and web pages, a weekly review to catch wrong or outdated memories, rules to keep me from over-relying on it or taking unchecked advice on health, money, or legal matters, a fallback plan if I switch tools or the service goes down, and a 30-day trial plan with measurable success criteria.
Recommended Tech Stack: An assistant or agent platform with memory controls, powered by a frontier model (Claude Sonnet 5.5, GPT-6 Sol, or Gemini 3.8 Flash), with connected apps (email, calendar, notes) given least-privilege access. Check each tool's memory, privacy, and deletion settings before connecting any account.
Quick Implementation Guide: Describe your role and daily routines, and ask the AI to state assumptions and list missing requirements. Set the memory rules, permissions, and approval list, then connect one app at a time. Run the 30-day trial, review remembered items every week, and delete anything wrong or outdated.
Real-World Use Case Example: Use this prompt to plan an always-on assistant that handles your morning brief, inbox triage, and follow-ups without acting on its own for risky tasks. Treat AI output as a starting point and verify permissions and memory behavior in the real tool.
Act as a designer of persistent personal AI assistants that remember context and act across apps. Design my always-on assistant setup for [role/work/life areas]. Include what it should remember and what it must never store (sensitive personal details and other people's private information), how I can view, edit, and delete its memory, which apps and data it can access with least-privilege permissions, tasks it may do alone versus tasks that need my approval (sending messages, spending money, deleting files), a daily routine (morning brief, inbox triage, task follow-ups), protection against prompt injection from emails and web pages, a weekly review to catch wrong or outdated memories, rules to keep me from over-relying on it or taking unchecked advice on health, money, or legal matters, a fallback plan if I switch tools or the service goes down, and a 30-day trial plan with measurable success criteria.
Recommended Tech Stack: An assistant or agent platform with memory controls, powered by a frontier model (Claude Sonnet 5.5, GPT-6 Sol, or Gemini 3.8 Flash), with connected apps (email, calendar, notes) given least-privilege access. Check each tool's memory, privacy, and deletion settings before connecting any account.
Quick Implementation Guide: Describe your role and daily routines, and ask the AI to state assumptions and list missing requirements. Set the memory rules, permissions, and approval list, then connect one app at a time. Run the 30-day trial, review remembered items every week, and delete anything wrong or outdated.
Real-World Use Case Example: Use this prompt to plan an always-on assistant that handles your morning brief, inbox triage, and follow-ups without acting on its own for risky tasks. Treat AI output as a starting point and verify permissions and memory behavior in the real tool.
