5 AI projects you can complete this weekend (vol. 2)

An AI voice receptionist

Intermediate, about 8 to 10 hours. A phone agent for a real local business type (dentist, salon, gym) that answers calls, handles the common questions, and captures booking details. The most visceral demo in AI right now: you hand someone a number and they talk to your build.

1. 🗣️ Build the agent in Vapi. Start from their assistant template, write the greeting and FAQ answers for your chosen business type, and pick an ElevenLabs voice. Vapi's starter credit covers all your testing.
2. 📅 Wire up bookings. Connect a Cal.com link so the agent can offer real appointment slots and capture name, number, and reason for the visit.
3. 📞 Give it a number. A Twilio number costs a couple of dollars a month and makes it a real phone line you can call live on camera or in a pitch.

I'm building an AI voice receptionist demo in Vapi for a [dentist /
salon / gym]. ElevenLabs voice, Cal.com for bookings, Twilio number.

1. Write the agent's system prompt: greeting, tone, the 10 most
   common caller questions for this business with good answers,
   booking flow, and how to hand off when it doesn't know.
2. Give me a test script of 8 caller scenarios, from easy to
   awkward, so I can stress-test it before I show anyone.
3. Write the 3-sentence pitch I'd send a real local business
   owner with the demo number.

A lead-gen enrichment engine

Intermediate, about 6 to 10 hours. An automation that scrapes a list of target businesses, has Claude research each one and write a personalized opening line, and outputs a ready-to-send outreach sheet. Lead enrichment is a literal line item in automation job listings, and its dollar value is obvious to any business owner.

1. 🕸️ Scrape the list. An Apify actor (Google Maps scraper for local, or a directory scraper for niches) pulls the raw list with names, sites, and contact details. Free credits cover a demo batch.
2. 🔎 Enrich each row. n8n loops the list through Claude: read the website, note one specific thing, and write a first line that proves a human-level look at their business.
3. 📤 Output the sheet. Everything lands in Google Sheets with columns for name, contact, insight, and the drafted line. That sheet is the demo.

Design my lead-gen enrichment workflow in n8n: Apify scrapes
[business type] in [city], Claude enriches each lead, output to
Google Sheets.

1. List the exact n8n nodes in order with their key settings.
2. Write the enrichment prompt: read the lead's website text and
   return one specific observation and a one-line personalized
   opener that references it, no generic flattery.
3. Tell me how to run a clean 25-lead demo batch without burning
   my free credits.

A Reddit lead-finder

Intermediate, about 6 to 8 hours. n8n watches chosen subreddits, Claude scores each new post for buying intent ("can anyone recommend...", "what do people use for..."), and high scorers land in a sheet with a Slack or email ping. Agencies charge four figures for this exact build.

1. 📡 Watch the subreddits. The Reddit node (or RSS feeds per subreddit) pulls new posts on a schedule into n8n.
2. 🎯 Score for intent. Claude rates each post 1 to 10 on buying intent for your chosen product or service, with a one-line reason. Only 7+ gets through.
3. 🔔 Ping the wins. Qualified leads land in a sheet and fire a notification, so the response happens while the thread is still warm.

I'm building a Reddit lead-finder in n8n for [product/service].

1. Give me the 8 subreddits most likely to contain buying-intent
   posts for this niche, with one line on each community's rules
   about promotion (I want to reply helpfully, not spam).
2. Write the intent-scoring prompt: rate a post 1-10 for buying
   intent, return the score, a one-line reason, and a suggested
   genuinely-helpful first reply.
3. List the n8n nodes in order to run this every 2 hours.

A content repurposing engine

Intermediate, about 8 to 10 hours. One long video goes in; transcripts, platform-native LinkedIn posts, tweets, and captions come out into a review queue. This is the system every creator and marketing team wants running internally, which makes it both a portfolio piece and a service.

1. 🎞️ Ingest the video. yt-dlp grabs the audio from any video URL; Whisper (free, local) or AssemblyAI's free tier transcribes it.
2. ✂️ Cut it into assets. Claude turns one transcript into five LinkedIn posts, ten tweets, and three reel captions, each written for its platform rather than copy-pasted.
3. 📋 Queue for review. Drafts land in a Google Sheet or Notion database with an approve column. Approved rows are ready to schedule. Never auto-post.

Design my content repurposing pipeline in n8n: video URL in,
platform-native drafts out to a review sheet.

1. List the pipeline stages and n8n nodes in order (yt-dlp,
   transcription, Claude, Sheets), with settings.
2. Write the repurposing prompt: from one transcript, produce
   5 LinkedIn posts, 10 tweets, and 3 reel captions, each
   following that platform's native style, no cross-posting voice.
3. Add a quality gate: how Claude should flag weak sections of
   the source video instead of forcing bad posts out of them.

A competitor intelligence digest

Intermediate, about 6 to 8 hours. A weekly automation that scrapes five competitors' public posts, has Claude rank what's working and why, and emails a briefing. It proves you can turn messy scraped data into decisions, which is the analysis skill clients actually buy.

1. 🗂️ Pick the five. Five competitor accounts or blogs in one niche. Public data only, and respect each platform's rules.
2. 📈 Scrape and score. An Apify actor pulls the week's posts with engagement counts; Claude ranks the outliers against each account's own baseline and writes the "why it worked".
3. 💌 Ship the briefing. One email, every Monday: top five posts, why they worked, and one suggested move. Scheduled in n8n, runs without you.

Design my weekly competitor digest: Apify scrapes 5 accounts'
posts + engagement, Claude analyzes, n8n emails me Monday 8am.

1. List the workflow nodes in order with settings.
2. Write the analysis prompt: rank posts by performance vs that
   account's own average (not raw numbers), explain the top 5 in
   one line each, and suggest one move I should copy this week.
3. Format the email so it reads in 60 seconds on a phone.

Turning a build into a paid gig

Every tool in one place

• Vapi (voice agents, $10 starter credit): vapi.ai
• ElevenLabs (voices, free tier): elevenlabs.io
• Twilio (real phone number, a few $/mo): twilio.com
• Cal.com (bookings, free tier): cal.com
• Apify (scraping, free monthly credits): apify.com
• n8n (automation, free self-host): n8n.io