50 AI Prompts Every Business Needs in 2026

I’ve built dozens of AI automations for clients, and I can tell you the single biggest waste of time is copying generic prompts from a listicle. The real value comes from prompts engineered as system instructions for your specific tools—Claude, GPT, n8n—that run on a schedule. Here are 50 prompts I’ve actually deployed, structured for you to plug into your own workflows in 2026.

Why generic prompts fail (and what to do instead)

You’ve seen them: “Write a catchy social media post.” Feed that to an API and you’ll get generic fluff. The failure happens because context is missing: your brand voice, your product specifics, your customer data, and the action you want the AI to take. A prompt isn’t a one-off query; it’s a config file for an automated process.

For example, a bad prompt is "Write a welcome email." A useful prompt is a system instruction for your Resend email workflow: ``json { "system": "You are a brand voice engine for [Your SaaS]. Use a friendly, technical tone. Always include one specific feature mention from the user's signup plan (stored in 'plan_type' variable). End with a clear, single CTA to book a onboarding call.", "user_data": "{{$json.user_email}}, {{$json.plan_type}}, {{$json.signup_source}}", "task": "Generate a personalized welcome email subject line and body." } `` This runs in an n8n workflow triggered by a new user in Supabase. The difference is specificity and data injection. Your prompts must be templates that pull from your database. Next step: Audit one manual task you do weekly. Document every piece of context you consider. Now, rebuild that as a structured prompt with placeholders for dynamic data. For a deeper dive on connecting these prompts to automations, see my guide on building AI employees.

10 Marketing & Content Prompts

These aren’t for brainstorming; they’re for feeding into scheduled content pipelines. Each assumes you’re pulling data from a CMS or analytics platform.

1. SEO-Cluster Article Brief: Act as an SEO strategist. Based on the primary keyword '{{keyword}}', generate a comprehensive article brief targeting the 'buyer intent' stage. Include: 5 LSI keywords, 3 subheadings (H2) as questions, and a meta description under 155 characters. Target word count: 1,800.
2. Performance-Post Mortem: Analyze the following post metrics: {{post_url}}, {{impressions}}, {{engagement_rate}}%. In 3 bullet points, diagnose why this under/over-performed. Then, generate 3 data-backed recommendations for the next piece in this series.
3. Social Carousel from Blog Post: Transform the key points from this blog post excerpt {{excerpt}} into a 5-slide LinkedIn carousel script. Slide 1: Hook. Slides 2-4: One insight per slide with a bold takeaway. Slide 5: Strong question CTA. Use concise, scannable text.
4. Personalized Cold Outreach (ABM): Using company data for {{company_name}} (industry: {{industry}}, recent news: {{news_snippet}}), draft a 3-sentence LinkedIn connection request. Focus on one specific challenge their industry faces and hint at a solution without pitching.
5. Product Update Video Script: Write a 60-second TikTok/Reels script announcing {{new_feature}}. Structure: Problem (3s), Solution Reveal (10s), Quick Demo (40s), CTA to learn more (7s). Use casual, excited language and include on-screen text cues in brackets.
6. Podcast Show Notes Engine: Convert the following transcript snippet {{transcript_chunk}} into show notes. Include: a 100-word summary, 5 timestamps with key topics, and 3 quotable tweet snippets.
7. Landing Page A/B Test Copy: Generate two distinct value proposition headlines for {{product_name}} that serves {{target_audience}}. Variant A: Benefit-driven. Variant B: Pain-point-avoidance. For each, write a supporting sub-headline.
8. Newsjacking Commentary: Based on this industry news article {{article_summary}}, draft a 250-word thought leadership post for our blog. Position our tool, {{product_name}}, as an enabler for the trend discussed. Maintain an authoritative, not salesy, tone.
9. Email Newsletter Curation: Review these 5 article links {{links_array}}. For our weekly newsletter 'The Stack', write a 2-sentence summary for each with a unique insight. Then, write a 100-word intro linking them all to a weekly theme.
10. Paid Ad Copy Variants: For the product {{product_name}} targeting {{audience_segment}}, write 5 distinct Google Ads headlines (30 chars max) and 2 description lines (90 chars max). Use these pain points: {{pain_point_1}}, {{pain_point_2}}. Include power words like 'automate', 'instant', 'effortless'.

Next Step: Pick prompt #1 or #3. Use n8n's Cron node to run it weekly, pulling a keyword from your Ahrefs/SEMrush API or a recent blog post from your Ghost/Strapi CMS. Output it directly to a Google Doc or Notion page for your writer. This turns AI prompts for business from a manual task into a content assembly line.

10 Sales & Lead Generation Prompts

These prompts qualify leads, personalize outreach, and handle early funnel conversations. They integrate with your CRM (like HubSpot or Pipedrive) and communication tools.

1. Lead Scoring Explanation: Review the lead's activity: {{page_views}}, {{demo_requested}}, {{company_size}}. On a scale of 1-10, score their readiness to buy. Provide a 2-sentence justification for the score and recommend the next immediate action for the SDR (e.g., "Send case study on X," "Call tomorrow AM").
2. Demo Call Personalization: Prepare a 3-point personalized agenda for a demo with {{contact_name}} at {{company_name}}. Use their website data {{website_snippet}} and stated challenge: "{{challenge}}". Connect each of our features {{feature_list}} directly to a point on their agenda.
3. Cold Email from Referral: You were referred by {{mutual_connection}}. Draft a 4-sentence cold email to {{prospect_name}}. Mention the referrer in the first line, establish a common ground (their work on {{project}}), and propose a single, low-lift next step (e.g., a 10-minute chat on topic X).
4. Post-Demo Follow-Up: Based on the demo call notes where the prospect showed interest in {{feature}} but was concerned about {{objection}}, write a follow-up email. Recap the discussed solution to their objection and attach a specific, relevant case study ({{case_study_title}}).
5. Pricing Page Chatbot Response: A visitor on the pricing page asks: "Is there a discount for annual billing?" Using our public pricing from {{pricing_page_url}}, craft a concise, friendly response that confirms the 20% annual discount, and asks a qualifying question: "Are you evaluating for a team or individual use?"
6. LinkedIn InMail for Event Follow-Up: We met at {{event_name}}. Write a personalized LinkedIn InMail to {{contact_name}} referencing our conversation about {{topic}}. Suggest a virtual coffee to explore a specific idea related to {{their_company_focus}} you had after the event.
7. Discovery Call Question Generator: For a prospect in the {{industry}} with the title {{title}}, generate 5-7 open-ended discovery questions that uncover budget, authority, need, and timeline (BANT) without being overly salesy. Focus on process and pain, not product.
8. Quote/Proposal Introduction: Using the agreed-upon scope items {{scope_items_array}}, write the opening "Summary of Understanding" paragraph for a proposal. Mirror the client's language from our calls and explicitly link the scope to their stated goals: {{client_goals}}.
9. Lost Lead Re-engagement: Six months ago, {{lead_name}} at {{company}} cited pricing as the reason for not buying. Draft a short email announcing a new, flexible "Starter" plan launched this quarter. Frame it as a direct response to past feedback.
10. CRM Activity Summary for Manager: Analyze the last 50 sales activities logged in the CRM. Identify the top 3 most common "next steps" and the average time between contact touchpoints. In 3 bullet points, provide one actionable recommendation to improve the sales process.

Next Step: Implement prompt #1 in an n8n workflow. Trigger it when a lead reaches a certain score in your CRM (via webhook). Have it post the score and justification to a dedicated Slack channel for your sales team. This turns raw data into an actionable alert. For more on orchestrating these triggers, check out my piece on n8n workflow templates.

10 Customer Support Prompts

These are for deflecting tickets, scaling personalized responses, and analyzing feedback. They plug into tools like Intercom, Help Scout, or a plain Slack channel.

1. Ticket Triage & Routing: Categorize this user request: "{{user_message}}". Possible categories: [Billing, Bug Report, Feature Request, How-To, Login Issue]. Also, assess urgency from 1 (low) to 3 (critical) based on language tone and problem severity. Output only JSON: {"category": "", "urgency": }
2. Personalized How-To Guide Response: The user with email {{user_email}} is asking how to achieve {{goal}}. They are on the {{user_plan}} plan. Using our knowledge base ({{knowledge_base_url}}), find the 2 most relevant articles. Summarize the key step from each in a single sentence, and provide the direct links. Acknowledge their plan limits/features if relevant.
3. Bug Report Summarization for GitHub: A non-technical user submitted a bug report: "{{user_description}}". Translate this into a structured GitHub issue. Include: a clear title, steps to reproduce (infer if missing), expected vs. actual behavior, and environment details (ask the user for app version, browser if not provided).
4. Empathetic Escalation Delay Message: We need to inform the user that their issue regarding {{issue_topic}} requires engineering investigation and will take 3-5 business days. Draft a transparent, apologetic, but confident message. Provide one tangible workaround if possible ({{workaround}}), and commit to a specific update timeline.
5. Feature Request Validation & Upvote Ask: A user suggested: "{{feature_request}}". First, validate the idea by paraphrasing the core user benefit. Then, draft a response thanking them, informing them they can upvote it on our public roadmap at {{roadmap_url}}, and asking one clarifying question to understand their use case deeper.
6. Refund Request Handling (Policy-Based): A user with account {{user_id}} is requesting a refund. Their subscription started on {{start_date}}, and they cited "{{reason}}". Our policy allows refunds within 30 days for technical issues. Analyze the dates and reason. Draft a response that either: a) politely grants the refund and outlines next steps, or b) politely declines, citing the policy, and offers an alternative (e.g., a month credit, help troubleshooting).
7. Support Conversation Summary: At the end of a resolved ticket, summarize the key issue ({{issue}}) and solution ({{solution}}) in 2 sentences. Format this summary for internal use in the CRM notes field on the user's account.
8. Churn Risk Analysis from Support Tickets: Analyze the last 5 support tickets for user {{user_id}}. Identify any recurring themes of frustration or feature gaps. On a scale of 1-10, how high is the churn risk? Provide a one-sentence rationale.
9. Proactive Downtime Notification: Our service is experiencing degraded performance for a subset of users regarding {{component}}. Draft a clear, non-technical status update for our public status page and Twitter. Acknowledge the issue, state we are investigating, and give a realistic time for the next update (e.g., "within the hour").
10. Feedback Survey Analysis Batch Job: Analyze 100 recent survey responses from {{survey_link}}. Identify the top 3 positive themes and the top 3 negative themes. For each negative theme, provide one concrete product or documentation improvement suggestion.

Next Step: Set up prompt #1 as an Intercom bot or an automatic first step in your n8n support workflow. Routing accuracy improves dramatically, and your team saves minutes per ticket. The key to effective AI prompts for business in support is structure—forcing a consistent JSON output that your automation can act on.

10 Operations & HR Prompts

Forget vague "improve efficiency" prompts. In operations and HR, you need prompts that plug directly into your systems and output structured data. I use these daily to automate workflows in n8n that connect to our HRIS or project management tools.

1. Meeting-to-Action Items: "Parse this meeting transcript from [Otter.ai export]. Extract every action item, assign it to the mentioned person, and format as a JSON array with keys task, owner, due_date (infer from phrases like 'by Friday' else default to +7 days)."
2. Policy Q&A Generator: "Using our employee handbook text [paste section], generate 5 multiple-choice quiz questions with 4 answers each (1 correct) to verify understanding. Output in JSON for our LMS."
3. Procurement Justification: "Draft a vendor justification memo for [Software Name]. Include: 1) Current problem (3 bullet points), 2) Cost analysis ($X/month vs. Y hours saved), 3) Security compliance check (SOC2, GDPR)."
4. Performance Review Draft: "Based on these 10 completed project tickets [paste titles/descriptions] and 3 peer feedback quotes [paste], draft three 'achievements' paragraphs and one 'area for growth' paragraph for a performance review."
5. SLA Breach Analysis: "Analyze this support ticket thread. Identify the exact timestamp of first contact, each handoff, and final resolution. Calculate delays. In one sentence, state the root cause of the SLA breach."
6. Interview Question Rubric: "For the role of [Senior Marketing Manager], create a scoring rubric (scale 1-5) for the question 'Describe a failed campaign.' Include criteria for strategic thinking, analytical depth, and accountability."
7. Contract Clause Simplifier: "Rewrite this legal clause from our Master Service Agreement into plain English for a client FAQ. Keep it under 100 words. [Paste clause]"
8. Onboarding Task Sequencer: "Create a 30-60-90 day onboarding plan for a [Remote SDR]. List 3 concrete tasks per week. Format as a table with columns: Week, Task, Success Metric, Required Resource."
9. Expense Report Audit: "Review this list of 15 expense line items [paste]. Flag any that violate our policy (no alcohol, >$75 requires receipt). Categorize each (travel, software, meals). Output as a CSV string."
10. Retrospective Synthesis: "Synthesize the 50 sticky note comments from our team retrospective in FigJam into 3 core themes. For each theme, propose one actionable experiment for the next sprint."

10 Developer & Technical Prompts

These are for building, debugging, and maintaining systems. I feed these directly into the Claude API or ChatGPT Code Interpreter. The key is constraining the output to a specific, usable format.

1. SQL Query from Description: "Write a PostgreSQL query for our users table (columns: id, email, created_at, plan_tier) and invoices table (id, user_id, amount, status, due_date). Find all users on 'pro' plan with at least one invoice overdue by >30 days. Return user email and total overdue amount."
2. n8n Workflow Node Builder: "Create an n8n function node in JavaScript that takes an input array of objects. For each object, if priority == 'high', add a field sla_deadline set to 2 hours from created_at. Return the modified array."
3. API Error Triage: "I'm getting a 429 error from the GitHub API. Here's my current logic: [paste code]. Suggest three specific mitigation strategies: 1) header inspection, 2) rate limit backoff logic, 3) cache implementation."
4. Log Analysis Pattern: "Scan this Vercel serverless function log [paste 50 lines]. Identify the most frequent error message. Hypothesize its root cause and suggest the exact code fix."
5. Supabase Migration Script: "Write a TypeScript script for Supabase Migration. It should: 1) Add a nullable onboarding_status column to public.profiles, 2) Backfill it as 'complete' for profiles with a last_login > '2024-01-01'."
6. Dockerfile Optimizer: "Optimize this Dockerfile for a Next.js app [paste]. Implement multi-stage build, correct layer caching for npm ci, and reduce final image size. Explain each change."
7. CLI Command Builder: "Write a zsh function to automate my deploy. It should: 1) Run npm test, 2) If passes, build for production, 3) Sync assets to S3 using awscli, 4) Trigger a Vercel deployment via their API."
8. Security Code Review: "Act as a security reviewer. Analyze this Express.js route for user upload [paste code]. List every potential vulnerability (e.g., path traversal, MIME type verification) and provide the fixed code snippet."
9. Observability Query: "Write a Datadog Logs Query to calculate the p95 latency of our POST /api/webhook endpoint over the last hour, grouped by status_code, only for errors."
10. Database Index Recommendation: "Given this slow query log entry [paste] and table schema [paste], recommend the exact CREATE INDEX statement to improve performance. Explain why you chose these columns."

How to customize these prompts for your industry

The prompts above are templates. To make them yours, you must inject your specific context, data schema, and tooling. Here’s my exact process.

First, replace the bracketed [variables] with real data. For the SQL prompt, that means pasting your actual table schemas from your Supabase or PostgreSQL \d output. For the operations prompts, use a real excerpt from your employee handbook or a past meeting transcript.

Second, constrain the output to your stack. If you use Linear instead of Jira, modify the prompt to output "Linear-compatible CSV." If you use Resend instead of SendGrid, specify the Resend API email object format. For example: "Format the output as a JavaScript array of objects compatible with resend.emails.send()."

Third, add your business rules. The procurement prompt needs your actual budget approval thresholds (e.g., "anything over $5k requires CFO sign-off"). The SLA analysis prompt needs your real SLA windows (e.g., "Tier 1: 2 hour response"). This turns a generic structure into an enforceable company policy.

I run these customized prompts through a Claude API workflow in n8n, storing the best-performing variants as templates in a database. The iteration is cheap, and the payoff is a library of company-specific automations that actually work.

Wrapping Up

Effective AI prompting isn't about magic phrases; it's about providing specific context, demanding structured outputs, and relentlessly iterating based on what works in your actual tools. Treat prompts like code—version them, test them, and refactor them. Start with one high-impact area, like turning sales call transcripts into structured CRM updates, and build your library from there.

Want all 500+ prompts organized by industry with variables and examples? Get the full prompt pack for $19.

Pick one prompt from this list, customize it with your data this week, and see what it builds for you.