Examples & Inspiration
Ideas, prompts, and starting points for your build.
What Could You Build?
Slack Feedback Bot
Monitors customer feedback, summarizes daily to Slack. Clustered topic analysis, top 5 issues, highlights.
CMO Agent "ChickenJoe"
Multi-source analytics (CRM, social, analytics), auto-posts content, identifies opportunities on its own schedule.
Meeting Notes Agent
Records meetings, extracts action items, and sends follow-up emails automatically to all participants.
Content Repurposer
Takes a blog post and generates social media posts, an email newsletter, and a Twitter/LinkedIn thread.
Invoice Processor
Watches an email inbox for incoming invoices, extracts amounts/dates/vendors, and adds them to a spreadsheet.
Hiring Pipeline Agent
Screens incoming applications, ranks candidates by fit, drafts personalized rejection/interview emails.
Pitch Coach
Voice-based startup pitch feedback. Pitch for 60 seconds, get spoken feedback on clarity, structure, persuasiveness.
AI Customer Service
Voice agent handling FAQs and objections for your product. Escalates to a human when needed.
Language Practice Partner
Conversational AI that chats with you in a foreign language, corrects mistakes, and adapts to your level.
AI Podcast Interviewer
A voice agent that interviews you about a topic, asks follow-up questions, and generates a polished transcript.
Competitor Analyzer
Paste two URLs, get a side-by-side comparison of positioning, messaging, and SEO gaps.
Sales Pitch Generator
Input a lead's URL and LinkedIn handle, get a tailored slide deck with AI-written copy and a shareable link.
Contract Reviewer
Upload a contract PDF, get a plain-English summary of key terms, risks, and unusual clauses highlighted.
Landing Page Generator
Describe your product in 3 sentences, get a complete landing page with copy, hero section, and CTA.
Prompt Masterclass
The difference between a mediocre result and something amazing is your prompt. Here's how to level up.
🎯 Rule #1: Be Ridiculously Specific
The more context you give, the better the output. Compare these two prompts:
❌ Vague
"Build me a chatbot"
✅ Specific
"Build me a customer support chatbot for a SaaS product. It should answer questions about pricing (3 tiers: Starter $29/mo, Pro $79/mo, Enterprise custom), handle cancellation requests by offering a 20% discount first, and escalate to email if it can't resolve in 3 messages. Use React for the frontend and Express for the backend."
📝 Rule #2: Show, Don't Tell (Few-Shot Prompting)
Instead of explaining what you want, show 2-3 examples. The AI learns the pattern instantly.
"Build me a tool that generates cold outreach emails. Here are 3 examples of the tone and style I want: Example 1: Subject: Quick question about [Company]'s growth plans Body: Hi [Name], I saw [Company] just raised Series A — congrats! We helped [Similar Company] cut their onboarding time by 40% after their raise. Worth a 15-min call this week? Example 2: Subject: [Name], loved your talk at [Event] Body: Hi [Name], your point about [Topic] really resonated. We're building exactly that — would love to show you a 2-min demo. Free Thursday? Example 3: Subject: How [Competitor] solved [Problem] Body: Hi [Name], not sure if you're facing [Problem] too, but [Competitor] cut their [Metric] by 60% using our approach. Happy to share the playbook — no strings attached. The tool should take a CSV with Name, Company, Recent News and generate personalized emails in this exact style."
🧠 Rule #3: Ask It to Think Step by Step
For complex tasks, tell the AI to break down its reasoning. This dramatically improves accuracy.
"Build a financial health dashboard for my SaaS business. Think step by step: 1. First, design the data model — what metrics do we need? (MRR, churn, LTV, CAC, runway) 2. Then build the API that calculates each metric from raw data 3. Then create a React dashboard with charts 4. Finally, add a monthly email report that summarizes trends Use Recharts for the graphs. The data comes from a PostgreSQL database with tables: subscriptions, payments, customers."
🎭 Rule #4: Give It a Role
Telling the AI WHO it is changes the quality of the output dramatically.
"You are a senior full-stack developer who has shipped 50+ production apps. You write clean, minimal code — no over- engineering, no unnecessary abstractions. You prefer simple solutions over clever ones. Build me a waitlist landing page with: - Email capture form with validation - Thank you page showing queue position - Admin dashboard to see all signups - Export to CSV button Use React, Tailwind, and a simple JSON file as the database (no need for a real DB for an MVP)."
Ready-to-Use Prompts
Copy, paste, adjust to your business. Each prompt is designed to produce a working prototype in one session.
🤖 Agent Track
"Build a Node.js agent that runs every morning at 8am and: 1. Reads the last 24h of entries from a Google Sheet (customer feedback column) 2. Groups them into themes using Claude API 3. Identifies the top 5 recurring issues 4. Generates a summary with: - Overall sentiment (positive/negative/neutral %) - Top issues with example quotes - One 'highlight of the day' (most positive feedback) - Suggested action items 5. Posts the summary to a Slack channel via webhook Use @anthropic-ai/sdk for the analysis. Include a CLAUDE.md with setup instructions. Make it deployable as a cron job."
"Build an agent that qualifies inbound leads:
1. Input: a CSV with columns [Name, Email, Company, Website, Message]
2. For each lead, the agent should:
a) Scrape the company website (just the homepage)
b) Analyze company size, industry, and what they do
c) Score the lead 1-10 based on these criteria:
- Company size > 50 employees: +3
- Industry matches [SaaS, FinTech, HealthTech]: +3
- Message mentions budget or timeline: +2
- Website has a pricing page (= real product): +2
d) Draft a personalized 3-sentence reply email
3. Output: a new CSV with all original data + score + draft email
4. Also print a summary: 'X hot leads, Y warm, Z cold'
Use Claude API for analysis and email drafting.
Use cheerio for lightweight HTML scraping." 🎙️ Voice Track
"Build a web app with a 'Record Pitch' button. When I click it: 1. Record my voice for 60 seconds (browser MediaRecorder API) 2. Transcribe the audio using OpenAI Whisper API 3. Send the transcript to Claude with this system prompt: 'You are a Y Combinator partner who has heard 10,000 pitches. Score this pitch 1-10 on: clarity, hook, problem definition, solution, and call-to-action. Be brutally honest but constructive. Give specific suggestions to improve.' 4. Convert Claude's feedback to speech using ElevenLabs API 5. Play the spoken feedback back to me Use React. Show the transcript and written feedback on screen while the audio plays. Include a 'Try Again' button."
"Build a voice-based AI receptionist for a dental practice.
The agent should:
1. Greet callers warmly: 'Hello, Dr. Müller's practice, how
can I help you?'
2. Handle these scenarios:
- Book appointment → ask for name, preferred date/time,
reason for visit, confirm
- Cancel/reschedule → ask for name, find booking, confirm
- Emergency → 'I'll connect you with the doctor immediately'
- Insurance questions → provide standard answer about
accepted insurances
- Anything else → 'Let me take your number and have someone
call you back'
3. Always be polite, never rush, confirm details before ending
Use ElevenLabs for voice synthesis, Whisper for recognition.
Build as a web app with a 'Call' button to simulate." 💻 Desktop App Track
"Build a competitive intelligence tool as a React web app. Flow: 1. User enters their website URL and up to 3 competitor URLs 2. The app scrapes all sites (homepage + /about + /pricing) 3. Claude analyzes and generates a report with: - Positioning comparison (who says what about themselves) - Pricing comparison (tiers, prices, features) - Messaging analysis (tone, target audience, key claims) - SEO basics (title tags, meta descriptions, H1s) - SWOT analysis for the user's site vs competitors 4. Display as a beautiful dashboard with tabs per section 5. 'Export as PDF' button Use cheerio for scraping, @anthropic-ai/sdk for analysis. Make it look professional — this is a tool founders will actually want to use again."
"Build a proposal generator web app for freelancers/agencies. Input form: - Client name and company - Project type (dropdown: Website, App, Branding, Marketing) - Brief description of what they need - Budget range - Timeline The app should generate a complete proposal with: 1. Executive summary (personalized to the client) 2. Scope of work (broken into phases) 3. Timeline with milestones 4. Pricing table (based on budget range) 5. Terms & conditions (standard template) 6. 'Why us' section Output as a beautiful HTML page with a 'Download as PDF' button and a 'Share via Link' option. Use Claude for all the writing. Make the generated proposals look so good the client thinks you spent 3 hours on it."
Pro Tips for Saturday
🏁 Start simple, then iterate
Get a basic version working in the first hour. Then improve it. Don't try to build the perfect thing on the first try.
💬 Talk to the AI like a colleague
Say "this doesn't look right, the button should be bigger" or "add error handling for when the API is down". Natural language works.
📸 Screenshot your errors
Claude Code can read screenshots. If something looks wrong, just paste a screenshot and say "fix this".
🆘 Use the support queue
Stuck for more than 10 minutes? Don't struggle alone. Go to support.buildyouragent.de and our tech team comes to your seat.
🔄 Commit often
Save your progress with git. If something breaks, you can always go back. Ask Claude Code to "commit my current progress".
🎯 Scope is everything
The #1 mistake: building too much. A working prototype that does one thing well beats a broken app that tries to do ten things.
The Golden Rule
Whatever you decide to build should excite you. The best results come from solving your own problems. Don't build what sounds impressive — build what you actually need.
