I built a podcast generator for myself earlier this year because I was terrible at reading research papers.

I'd subscribe to arXiv updates in AI, crypto, and quantum computing. Save papers. Never read them. Meanwhile, I listened to podcasts 2-3 hours daily during commutes and workouts.

So I built a simple pipeline: feed in papers, GPT-4 extracts key points, ElevenLabs creates audio, drops into my podcast feed. Just for me.

It worked surprisingly well. I was finally staying current on research—while doing laundry.

In September, I met a patent litigation attorney with 25 years of Federal Circuit experience.

I showed him my research podcast tool. He got excited.

"Patent attorneys have the exact same problem," he said. "Federal Circuit publishes 100+ opinions per year. We need to stay current, but reading full opinions takes hours we don't have."

Then he said something that changed everything:

"We also need 12-15 hours of Continuing Legal Education every year. It's mandatory."

Mandatory.

That word made it all click.

Researchers want to stay current on papers. It's optional.

Patent attorneys must earn CLE credits. The State Bar requires it.

Current CLE options suck:

• $800-1,500 conferences

• Full days blocked from billable work ($400-500/hour lost)

• Content 6+ months outdated

• Generic topics

But attorneys are already commuting, working out, doing tasks that don't need full focus—perfect for audio.

What if we collapsed two obligations into one?

Listen during commute → Stay current on Federal Circuit → Earn CLE credits.

All in 30-minute weekly episodes.

I rebuilt everything around this focus:

The content: Federal Circuit patent law opinions, analyzed by a domain expert, produced using my AI pipeline

The infrastructure: 2,000+ commits to build CLE-compliant systems

• Verification tracking (who listened, how long)

• Certificate generation (State Bar requirements)

• Transcript creation (written materials required)

• User dashboards (progress, deadlines)

The business model: $100/month for CLE-accredited professional education (vs. $10/month research subscriptions)

The go-to-market: Patent attorneys with allocated CLE budgets and mandatory requirements (vs. academics with no budgets)

November 2025:

• 120+ on waitlist

• 15+ podcast episodes produced

• Filing California CLE accreditation (expected Q1 2026)

• Beta testing with top law firm: "Exceptionally well done"

• Zero revenue (waiting for accreditation to start billing)

The plan:

• Launch with 45K patent attorneys

• Expand to 10 legal verticals (securities, tax, family law)

• Scale to all licensed professions (CPAs, doctors, architects)

• Add video content alongside audio—becoming the Netflix for professional education

• 15M+ professionals need continuing education annually

Why video matters: Some concepts need visuals. Patent claim diagrams. Case timelines. Financial statement analysis. Surgical procedures. We're building both formats—audio for commutes, video for deep dives. Same AI pipeline, same compliance infrastructure, same seamless experience.

How we'll execute:

Phase 1: Audio Foundation (2026) Master the podcast format first. Get California CLE accredited. Prove attorneys will pay $100/month for audio that earns credits. Build subscriber base of 170-350 patent attorneys. Perfect the compliance infrastructure—verification, certificates, transcripts.

Phase 2: Video Pilots (Late 2026) Start with topics where video adds clear value. Patent claim construction tutorials. Federal Circuit oral arguments with synchronized transcripts. Case law timelines with visual flowcharts. Test with 10-20 subscribers. Gather feedback. Refine production pipeline.

Phase 3: AI-Generated Video (2027) This is where it gets interesting. We already have:

• Research showing AI can generate educational animations (my SIGCSE '24 paper on dynamic graph generation)

• Tools like Manim for programmatic video creation

• LLMs that can create professional animations without fine-tuning (proven by 3Blue1Brown)

The pipeline: Legal case → GPT-4 extracts key points + visual elements → Manim generates animations → ElevenLabs for voiceover → Final video output.

Same automation that makes our podcasts scalable, now applied to video. We can produce 10x more content than traditional CLE providers at 1/10th the cost.

Phase 4: Hybrid Content Library (2027-2028) Every topic gets both formats:

• Quick audio version (30 min, commute-friendly)

• Deep dive video (60-90 min, with visuals, diagrams, examples)

• Interactive quizzes (optional, for retention)

• AI tutor (chat with questions about the content)

Users choose based on context: Audio on the train. Video at home. Quiz before exam. Tutor when confused.

Phase 5: Creator Platform (2028+) Enable other attorneys to create accredited content on our platform. We provide:

• Production tools (AI pipeline for audio/video)

• Compliance infrastructure (verification, certificates)

• Distribution (our subscriber base)

• Monetization (70% to creator, 30% to us)

Suddenly we're not limited by how much content we can produce. We become YouTube for professional education—except every video earns you credits.

One platform. Multiple learning modes. Professional education that finally fits how people actually consume content in 2025.

The tool I built for myself wasn't the product.

It was the prototype for understanding what works.

Researchers: Optional learning, no budget
Patent attorneys: Mandatory compliance, allocated budgets

Free content: Nice to have
CLE credits: Legal requirement

Generalized platform: Hard to get traction
Hyper-specialized niche: Clear value proposition

The research paper podcast solved my problem.

Research Club solves a problem people are required to solve—and makes it feel like learning instead of compliance.

Started with me wanting to keep up with arXiv papers during my commute.

Evolved into transforming how 15 million licensed professionals fulfill mandatory education requirements.

From personal hack to professional platform.

From audio-only to full Netflix-style content library.

From single producer to creator marketplace.

From solving my problem to building the infrastructure that makes professional learning continuous, personalized, and compliant.

The research paper podcast I built in January wasn't just a solution to my problem. It was proof that AI can make educational content production 10x cheaper and 10x more personalized than traditional methods.

Now we're taking that insight and applying it to an industry desperate for change—where 15 million people spend $8-10 billion annually on education they're required to get but don't enjoy getting.

The playbook is clear. The market is massive. The unit economics work. The technology exists.

Now it's execution.

That's the journey.