I’m building the tool Harry Stebbings said he’d pay $100k for.

The backstory

I’m Nelson. I’ve built and raised for two VC-backed companies, so I’ve spent a lot of time around VC podcasts — as a guest, as a listener, and as someone who kept noticing the same thing: incredible content getting posted once and forgotten about.

The pattern is always the same. Host records a 45-minute conversation with a brilliant guest. Posts the full episode. Maybe grabs one clip for social. Moves on to the next recording. Meanwhile, 5–8 genuinely shareable moments just... evaporate.

Your latest podcast episode isn’t your best content. It’s just your newest. God knows how many killer 30-second segments are sitting in dusty archives generating absolutely zero value for anyone.

Why I didn’t just use Opus Clip

I tried every AI clipping tool on the market. Same fundamental problem with all of them: they’re trained on what goes viral on TikTok. High energy. Emotional outbursts. Rapid-fire takes.

But the moments that actually get shared in VC circles are completely different. A GP quietly explaining why they passed on a deal that became a unicorn. A founder describing the 18 months before product-market fit. A contrarian take on valuations that makes someone stop scrolling on LinkedIn.

Low energy, high signal. Generic tools miss these every time.

So I built my own

Step 1 was “can AI match a top editorial team’s taste in clip selection?”

I started with 20VC, the world’s largest VC podcast. Analysed 870+ editorial decisions — which moments Harry’s team chose to clip, what they posted, what they didn’t. The goal was to build a system that could replicate those decisions.

Results were hit and miss at first (83% on one episode, 21% on the next), until I realised that if I first classified the type of episode, results got way better. Good learning.

In a blind test, Clypt now matches 83% of the 20VC editorial team’s clip selections. Not by mimicking their style, but by learning the patterns behind which moments actually resonate in VC.

Step 2: what actually performs on social

Finding the right clips is only half the battle. The other half is knowing how to frame them for social. So I pulled engagement data from thousands of clip posts across X and LinkedIn, figured out what patterns drive shares vs. what just generates impressions, and baked all of that into the system.

Something I’m mindful of: engagement data can be noisy. One big account retweeting can make an average clip look amazing. So the model weights editorial quality higher than raw engagement. The goal is clips your audience genuinely finds valuable, not just ones that game the algorithm.

What you actually get

You send me an episode. Within 48 hours, you get back 5–8 ranked clips. Each one includes:

• A clip archetype — Counterintuitive Take, High-Stakes Story, Bold Claim, Vulnerable Moment, or Tactical Playbook. (The 5 archetypes)
• Editorial rationale — why this moment works as a standalone clip
• A scroll-stopping hook — the first line of the social caption, optimised for LinkedIn and X
• Ready-to-post copy — caption, context, and CTA for each platform

You review, approve, and post. One episode becomes 2–3 weeks of social content.

Who this is for (and who it isn’t)

Clypt is built for VC podcast hosts who know their content is valuable but don’t have the time or team to extract every shareable moment from every episode. If you’re a GP running a podcast alongside managing a fund, or a content lead at a VC firm trying to get more from every recording hour, this is for you.

I’m not building a tool for every podcaster. Deliberately. If your show is about cooking or true crime, we’re probably not the right fit. VC conversations have different rhythms, different audiences, and different measures of success. That specificity is the whole point.

Where this is going

Right now I’m onboarding the first few shows and building in public as I go. The model gets better with every podcast I work with — more editorial decisions, more patterns, better clip selection.

If you run a VC podcast and want to stop manually scrubbing through hours of footage to find your best clips, I’d love to chat.