Published
04/03/25
Share

Synthesia's $2B+ Secrets: Finding Product Market Fit & Avoiding Hype in AI

I sat down last week at our GenAI Meetup to talk to Synthesia’s founder Victor Riparbellii. I’ve been fortunate enough to have witnessed Synthesia’s growth over the last 5 years. Victor shared brutally honest lessons learned navigating the hype, finding product-market fit, and building an enduring AI company.

Below is his unfiltered advice for founders and investors on navigating the AI hype.

The text is vibe-scribed by Gemini 2.0 Advanced (with some light human editing) from a transcript of our talk.

1. Find the real problem (hint: it’s not always the obvious one)

Synthesia didn’t start by targeting corporate videos. Their initial bet? AI dubbing for Hollywood and ad agencies. Makes sense, right? Video experts using video tech. It didn’t work out. “$1M revenue in 18 months… it was a vitamin, not a painkiller,” Victor admits. The quality wasn’t there, and the real need was missing.

The pivot came from listening to who was desperately contacting them: corporate folks stuck with text and slides, unable to get budget or skills for real video. Lesson: Don’t assume your revolutionary tech fits existing expert workflows. Look for the unmet need, even if it seems less glamorous.

2. Your competitor isn’t always who you think (embrace “good enough”)

Everyone compared Synthesia’s early avatars to highly produced videos. “Compared to that, it was crap,” Victor stated bluntly. But their actual users weren’t comparing it to Hollywood. They were comparing it to a 20-page PDF for onboarding fast food staff or a text-based support doc.

Suddenly, Synthesia’s videos weren’t just good enough; they were way better than the alternative. Lesson: Find the niche where the current alternative sucks more than your imperfect solution. Your true competitor might be a PDF, not Pixar. That “good enough” market can be massive and was Synthesia’s “secret for many years.”

3. Build a workflow, not just a wrapper

In the AI gold rush, it’s tempting to slap a UI on a cool model and call it a product. Victor argued that’s a recipe for zero defensibility. Synthesia won by understanding the entire user workflow. Their users weren’t video editors; they were PowerPoint users. So, they built a familiar, PowerPoint-style editor.

They didn’t stop there. They added collaboration, content management, translation, publishing, even their own video player. “The model is a part of a bigger whole.” Lesson: Think end-to-end. Build the platform around the AI magic. That creates lock-in and defensibility that a thin wrapper can’t match. Customers care about workflows, not just model benchmarks.

4. Tune up your “BS filter” (avoid innovation potholes)

“You get taken in by the hype… build things that if you really ask ‘why’ six times… the answer’s maybe no.” Victor calls these “innovation potholes.” Examples abound: chasing early personalized sales videos that clearly didn’t work, or nearly getting sidetracked by a massive consulting firm’s doomed metaverse project (“We just didn’t believe”).

Lesson: Be ruthless about filtering hype, vanity metrics, and even lucrative deals that distract from your core belief. Ask “why”. Does it really solve a problem? Does it really work? Don’t divert precious resources chasing trends or novelty.

5. Remember the boring stuff wins

With tiny teams building $5M ARR products in months using LLMs, it’s easy to think the old rules don’t apply. Victor disagrees. While AI accelerates things, enduring companies aren’t built just on tech magic.

He finds himself reading less about AI and more about classic SaaS go-to-market strategies from a decade ago. “Good go-to-market, being super customer-driven… those are enduring business lessons.” Especially in a crowded market, building relationships, understanding customers, and executing classic SaaS plays will differentiate winners. Lesson: AI is a powerful tool, but fundamentals still matter. Don’t neglect GTM, sales, and real customer connection amidst the tech frenzy.