Hey everybody, welcome back to PMF 🔥
ChatGPT usage chart was leaked, and it’s crazy:
It’s mostly used for the following activities:
Tutoring or teaching
Knowledge
How to Advice
Health, Fitness, Beauty or SelfCare
Creative Ideation
Personal Writing or Communication
Edit or critique text
Translation
Summary Generation
Each category is a signal of real demand and a blueprint for billion-dollar companies.
So I thought it would be really interesting to dig deep into each use case and come up with a business idea you can build on top of chatGPT (and knowing you already have the demand for it)
Framer is giving early-stage startups a full year of the Pro Plan: completely free.
Why Framer?
Framer is the design-first web builder that helps startups launch fast. Join hundreds of YC founders already building on Framer.
Who’s Eligible?
Pre-seed and seed-stage startups.
Offer available to new Framer users only.
What You Get:
✅ 1 year of Framer’s Pro Plan ($360 value)
✅ Fast, professional site — no dev team needed
✅ Scales with your startup — from MVP to full site
Spots are limited — apply before October 31st to claim your free year.
If you’re a top tech company and want to help thousands of founders and investors grow. Sponsor PMF at g@guillermoflor.com
The 17 billion-dollar AI startup ideas based on chatGPT Usage:
1) AI Personal Tutor Marketplace
Product: Marketplace that matches learners with AI tutors fine-tuned for specific exams/skills, with progress tracking, assignments, and accountability. Think: AI tutor + human optional review layer.
GTM: Start with one exam where outcomes matter (SAT/GMAT, IELTS, USMLE, bar prep). Win by being “the highest score increase per hour.”
Difficulty: Commoditized “chat tutoring” is easy. Differentiation comes from curriculum, diagnostics, and outcome proof.
How to start: Ship one vertical: diagnostic test → study plan → daily drills → score prediction. Layer marketplace later (top tutors contribute lesson plans, get paid per completion).
Moat: Proprietary learning data + high-trust brand around outcomes.
Example: Tutor AI
2) Corporate Upskilling Copilot (start with SDR onboarding)
“Reduce SDR ramp time by 30% using a copilot trained on your playbook.”
Product: Internal copilot that teaches employees company-specific tools and processes: “How do we do X here?” plus guided practice.
GTM: Sell to L&D and team leads with a tight ROI story: reduce ramp time for new hires and cut repetitive manager questions.
Difficulty: Enterprise procurement + data security. You need admin controls, audit logs, and content governance.
How to start:
List 50 startups hiring SDRs (LinkedIn jobs filter).
Email Head of Sales/RevOps: “I can build an SDR onboarding copilot in 48h using your docs.”
Offer: “$500 pilot for one team, cancel anytime.”
Deliver as Slack bot + Notion Q&A with citations (manual backend ok).
Measure: # questions answered, time saved, ramp milestones.
Turn the pilot into $2k–$5k/mo.
Moat: Deep integration into internal knowledge + usage-based lock-in as it becomes “the way you onboard.”
3) DIY AI Coach for repairs, cooking, and crafts (start with 1 object)
Product: Visual, step-by-step coach that uses the camera, detects where you are in the task, and prevents mistakes.
GTM: Go where DIY already lives: YouTube creators, Home Depot-style communities, Reddit niches, TikTok DIY.
Difficulty: Real-world edge cases. You need narrow scopes to avoid hallucination-risk moments.
How to start:
Pick one niche with obsessed users (espresso, bikes, 3D printers).
Scrape top 50 recurring problems from Reddit/forums.
Create 10 guided flows with images + decision tree.
DM posters who asked those questions: “I built a step-by-step coach—want to try it?”
Concierge: they send video/photos; you respond with steps + links.
Charge $29/mo or $9 per fix. Collect before/after photos.
Moat: A library of validated procedures + partnerships with tool brands and retailers.
Example: DIY-AI Fix-it
4) B2B SOP Generator that turns messy processes into playbooks
“Turn your messy process into SOP + checklist + training quiz in 48h.”
Product: Record a workflow (calls, screens, docs) → auto-generate SOPs, checklists, training, and QA rubrics.
GTM: Mid-market ops teams drowning in tribal knowledge: agencies, BPOs, logistics, healthcare admin.
Difficulty: Trust and adoption. SOPs fail when they’re inaccurate or ignored.
How to start:
Target agencies/BPOs/ops-heavy SMBs (search “operations manager” + “agency”).
Cold email: “Send me a Loom of the process, I’ll return an SOP + quiz in 48h.”
Offer: $300 for first SOP. Bundle 10 SOPs for $2k.
Deliver via Notion + Google Doc + Loom walkthrough.
Ask: “Which process should I do next?” (this reveals roadmap)
Upsell monthly retainer to document all processes.
Moat: Process graph of how the company runs. Once you own that, you become core infrastructure.
Example: Scribe
5) Vertical AI Search Engines for law, medicine, or finance
“Draft a legal memo with citations in 10 minutes.”
Product: Search with source-grounded answers, citations, and jurisdiction/company context. Not “chat,” but “decision-grade retrieval.”
GTM: Win one niche where speed matters and mistakes are expensive: immigration law, oncology literature triage, SEC filings analysis.
Difficulty: Accuracy, liability, and dataset licensing.
How to start:
Pick one corpus you can access (public case law / SEC filings / PubMed).
Create a Chrome extension or simple web UI that returns answers + citations.
DM 50 professionals: “I made a faster way to search X with cited answers. Want access?”
Give 10 people free access for feedback, schedule 15-min calls.
Watch them use it on screen. Fix the top 5 failure modes.
Convert to paid: $49–$199/mo depending on role.
Moat: Domain-specific ontology + feedback loops from expert users + licensed data.
6) Real-time Fact-Checked Copilot
“Flags risky claims in your doc and suggests sourced replacements.”
Product: A writing/browsing copilot that flags claims as you type and suggests verifiable replacements with sources.
GTM: Journalists, researchers, analysts, and anyone publishing. Start with teams where credibility is the product.
Difficulty: Defining “truth” and handling ambiguous claims. Needs transparent confidence and sourcing.
How to start:
Build a Google Docs add-on / browser extension (or manual: upload doc).
Target 50 newsletter writers/journalists on X/LinkedIn.
Offer: “I’ll fact-risk scan your next piece for free.”
Deliver annotated doc: claim highlights + sources.
Ask them to post “fact-checked by X” as credit.
Productize into subscription + team plan.
Moat: Claim graph + publisher integrations + reputation as the safest layer.
Example: Originality.ai
7) Enterprise Writing Copilot that integrates with Salesforce, HubSpot, or Jira
Product: Writing that is context-aware: it knows the account, ticket history, deal stage, and past comms.
GTM: Sell to RevOps, Support Ops, and CX leaders. Position it as “faster throughput + consistent tone + fewer mistakes.”
Difficulty: Integration depth and permissions. Shallow “text generation” will lose.
How to start: One integration, one job: “reply to tickets” or “write follow-ups” pulling only approved fields.
Moat: Embedded workflows + templates tuned to company outcomes.
8) AI Ghostwriter for founders and executives
“Turn your weekly voice note into 10 posts in your voice.”
Product: A personal content engine: extracts voice, ideas, and POV, then turns them into posts, emails, memos, and speeches.
GTM: Top-down distribution. Sell to people with audiences or internal leverage (founders, VPs, partners).
Difficulty: Voice authenticity and trust. If it feels fake, it dies.
How to start: A “weekly interview” workflow: 20-minute voice note → 10 draft outputs → user selects and edits.
Moat: Proprietary voice model + idea graph + performance learning from engagement.
Example: Imagine AI
9) AI Editor Pro for journalists and authors
Product: Developmental editing, structure, pacing, argument clarity, and line edits. Not grammar. Real editorial brain.
GTM: Authors, newsletters, editorial teams, agencies. Start with paid creators who ship weekly.
Difficulty: Editors are picky. You need controls, not magic.
How to start: Offer “editor modes”: tighten, restructure, fact-risk scan, tone consistency, plus an edit history view.
Moat: Personalized editorial preferences + dataset of what edits lead to better outcomes.
Example: HyperWrite
10) Student Writing Assistant
Product: Teaches writing while preventing lazy cheating. Suggests outlines, thesis, counterarguments, sources, and revision plans.
GTM: Schools and parents. Also direct-to-student with “grades improved” loops.
Difficulty: Academic integrity and policy pressure.
How to start: Build “process enforcement”: you must submit notes, outline, draft, revision. Reward learning steps.
Moat: School partnerships + alignment with integrity requirements.
Example: https://jenni.ai/
11) AI Relationship Coach
Product: Conversation prep, conflict resolution scripts, attachment-style insights, and “post-argument repair” plans.
GTM: DTC subscription. Influencer distribution is strong here. Also couples therapists as a channel.
Difficulty: Safety and overreach. Must avoid pretending to be a therapist.
How to start: Narrow to one scenario: “fight recovery” or “hard conversations.” Provide structured prompts, not diagnoses.
Moat: Habit + personalization + trusted brand.
Example: https://textmei.com/relationship-ai/
12) Job Application Engine for resumes and cover letters
Product: ATS-optimized resume tailoring, cover letter generation, recruiter outreach, and interview prep, all linked to job descriptions.
GTM: Viral DTC + partnerships with bootcamps/universities.
Difficulty: Crowded market. You need a sharper wedge than “AI resume.”
How to start: Own one segment: new grads in tech, nurses, consultants, product managers. Include measurable outputs: interview rate lift.
Moat: Outcome tracking + employer-side signals + proprietary templates that actually convert.
Example: aiApply
13) AI Personal Health Companion
Product: Helps users understand labs, meds, symptoms, lifestyle experiments, and questions for doctors, with strict disclaimers.
GTM: DTC subscription, employer wellness, partnerships with labs and telehealth.
Difficulty: Regulation-adjacent trust. Must be conservative and source-based.
How to start: Choose one “safe” wedge: lab interpretation + question generation for physician visits.
Moat: Longitudinal personal health history + clinician-in-the-loop pathways.
14) AI Mental Wellness Coach
Product: Guided journaling, CBT-style exercises, habit tracking, and escalation pathways.
GTM: DTC + employer benefits + therapist referrals.
Difficulty: Safety. Must handle crisis cases responsibly and avoid false authority.
How to start: Non-clinical, evidence-informed routines: sleep, anxiety journaling, thought reframes, accountability.
Moat: Daily habit + personalized playbooks + partnerships with care networks.
15) Full-stack AI Dev Shops that build MVPs end to end
“MVP in 21 days: landing, auth, payments, admin.”
Product: “Agency 2.0” where AI does 70% and humans do architecture, product judgment, and shipping. Fast MVPs with predictable scope.
GTM: Founder networks, accelerators, SMEs. Strong demand when hiring is hard.
Difficulty: Scaling delivery and maintaining quality. Agencies break when people become the bottleneck.
How to start:
Post a fixed offer on LinkedIn daily for 7 days.
DM 50 founders raising pre-seed.
Offer: $5k–$15k fixed scope. Clear deliverables.
Deliver fast with templates.
Ask for referral to 2 founder friends.
Scale via cohorts and repeatable stacks.
Moat: Reusable internal frameworks + distribution + reputation.
16) AI Ad Creative Studio for video, copy, and images
“50 ad variations/week + testing plan.”
Product: Produces ad variants at scale, predicts performance, and ties into campaigns. The real product is iteration velocity.
GTM: Performance marketers and DTC brands. Also agencies.
Difficulty: Everyone can generate creatives. Winning is closed-loop learning with results.
How to start:
Target DTC brands spending $10k+/mo on Meta.
Cold email: “Send 3 winners, I’ll generate 30 variants in your brand.”
Deliver in a folder + spreadsheet + naming conventions.
Charge $500 pilot, then $2k–$10k/mo.
Pull results back to learn what works.
Build performance loop as moat.
Moat: Creative-performance dataset and brand-specific learnings.
Example: Adcreative.ai
17) AI Storyboarding Studio for film, gaming, or content creators
Product: Script → storyboard → shot list → animatic. Integrates with production tools and style libraries.
GTM: Creators, studios, indie game teams, YouTubers. Start where budgets are tight but output is constant.
Difficulty: Taste and workflow fit. Artists resist tools that feel generic.
How to start: One creator persona: YouTube explainer channels or indie short films. Give them “storyboard in their style” quickly.
Moat: Style consistency + creator libraries + collaboration features.
Example: https://ltx.studio/
So, I thought this was interesting, and it’s cool to see many of the ideas already have players. However there are thousands of spaces to fill.
I’ll probably dig deeper into this and do this exercise better and maybe choose one idea to build myself.
Hope you found it valuable, and if you did, don’t forget to share and subscribe!
Cheers,
Guillermo