EO StudioThe 8-Year Journey to Building a $4B AI Company | Synthesia, Victor Riparbelli
CHAPTERS
- 0:00 – 1:02
Synthesia at $100M ARR: usefulness over hype in AI video
Victor opens by grounding Synthesia’s scale—$100M ARR and rapid early growth—then immediately frames the core product lesson: novelty isn’t enough. The chapter sets up the central tension for AI products today: viral curiosity vs recurring customer value.
- •Synthesia’s revenue milestones and rapid ramp from $1M ARR onward
- •The key question: is the product “fun/new” or genuinely useful?
- •Customers buy outcomes (solving a problem), not models
- •Viral attention can mask weak retention
- •Company perception swings between “hot” and “incumbent,” but value must stay constant
- 1:02 – 1:32
What Synthesia does and who it serves
Victor formally introduces himself and Synthesia’s mission: making video creation accessible and effective for business communication. He positions the product as a “PowerPoint-like” workflow for creating engaging videos at scale.
- •Synthesia is an AI video platform focused on enterprise/business use
- •Goal: help customers deliver messages with maximum engagement
- •Product framing: anyone who can use PowerPoint can make a video
- •Scale signals: 70% of Fortune 100, 65,000+ customers
- •Business traction: $100M ARR and significant venture backing
- 1:32 – 2:33
The hard early years (2017–2020) and the technical uphill battle
He describes the first three years as difficult: the tech was immature and progress was slow. This chapter highlights how early breakthroughs in neural video generation created conviction even before a clear product existed.
- •Company started in 2017; first years were not “fun” and tech didn’t work well
- •Early neural nets could generate video in real time at surprising quality
- •Comparable output would have cost millions via traditional VFX
- •Breakthrough triggered a belief that video creation would fundamentally change
- •Conviction preceded clarity on the initial use case
- 2:33 – 3:03
Optimism as an edge: betting on technology during the AI winter
Victor argues that people over-index on how new technologies can go wrong, while most usage trends positive. He frames optimism—especially early—as a strategic arbitrage when the market is skeptical.
- •Default human reaction to new tech is fear and failure modes
- •Most technologies are mostly used for good, with a minority of misuse
- •Being optimistic early can be a competitive advantage
- •Synthesia pitched a bold future: Hollywood-level creation from a laptop
- •The company began during an “AI winter” with widespread disillusionment
- 3:03 – 4:04
Getting rejected by ~100 investors—and the one “yes” that mattered
He recounts prolonged fundraising difficulty until a cold email to Mark Cuban connected with a shared vision. The lesson: find believers who already share the future you see, because you only need one committed backer to start moving.
- •Turned down by roughly 100 investors before traction
- •Mark Cuban aligned with the long-term vision; evaluated team execution instead
- •Fundraising lesson: one yes is enough
- •Easier to find investors who share your worldview than to convert skeptics
- •If you’re deeply convinced the opportunity will exist, persistence can be rational
- 4:04 – 4:34
Early roadmap triage: what can we build and sell with limited capital?
With only about $1M to start, Synthesia had to choose a narrow, fundable path that could demonstrate progress before the next raise. Victor explains the difficulty of deciding when to persist versus pivot when the vision is clear but the product path isn’t.
- •Constraint-driven planning: build and sell something before needing the next round
- •Tension: conviction in tech value vs uncertainty about first market entry
- •Knowing when to continue vs pivot is one of the hardest founder calls
- •Early execution doubt is normal even with strong market belief
- •The initial product had to be both feasible and sellable quickly
- 4:34 – 5:34
The first attempt: AI dubbing as a ‘vitamin,’ not a ‘painkiller’
Their initial use case—AI dubbing with facial reanimation—had real theoretical value but couldn’t scale due to quality demands and technical limitations. It became a classic “cool demo” that customers liked but didn’t truly need.
- •AI dubbing concept: translate video by changing voice + reanimating lips/face
- •Early limitations: required straight-on camera; heavy manual effort; slow output
- •Film/production customers demand extremely high quality thresholds
- •Insight: valuable in theory, but not urgent enough to drive dependency
- •Conclusion: it was a “vitamin,” not a “painkiller” (no one would miss it)
- 5:34 – 6:35
Finding the real market: enabling people who don’t make videos (corporate comms)
Customer discovery revealed a massive segment: people who should communicate via video but lack skills, budget, or editability. Synthesia narrowed the scope to corporate talking-head communication—simpler than Hollywood—and built avatar tech that unlocked real adoption.
- •Big unmet need: billions who want video but can’t produce it
- •Pain points: no camera skills, no budget, editing is hard after recording
- •Corporate videos are easier than cinematic content and more repeatable
- •Narrowing the domain made the problem tractable and valuable
- •Avatar technology became the core product and catalyzed takeoff
- 6:35 – 7:05
Product-market fit as art: combining customer insight with technical reality
Victor emphasizes that PMF wasn’t obvious from day one; it emerged as the team learned what was technically possible and what customers truly valued. He argues that “we should’ve built it earlier” ignores the learning curve and evolving research.
- •High conviction in the tech didn’t equal clarity on the first use case
- •PMF required iterative learning and close customer understanding
- •Research progress over time enabled the avatar approach
- •Retrospective “should’ve done X on day one” often unrealistic
- •Successful direction came from blending customer needs + tech insight
- 7:05 – 7:35
Viral launch vs retention: separating ‘tourists’ from recurring users
After the avatar MVP launched, Synthesia went viral—but most users were curious experimenters who didn’t return. The team focused on the smaller cohort with repeat behavior to understand the durable value proposition.
- •Viral growth brought many users creating demo videos for fun
- •Retention revealed most didn’t have ongoing needs
- •A smaller group kept returning and building regularly
- •Deep dives with repeat users uncovered the real value drivers
- •AI lesson: inbound hype can distract from identifying true recurring use cases
- 7:35 – 9:06
The key positioning shift: customers compared Synthesia to documents, not films
Synthesia’s “quality bar” became workable once the team understood the true alternative customers were replacing. For many enterprise users, the baseline wasn’t professional video—it was text documents—changing feature priorities and expectations.
- •Powerful insight: comparison set was text docs, not camera/video production
- •Lowered the required realism threshold for business use cases
- •Changed which features mattered most to users
- •Customer-guided iteration requires critical thinking amid high inbound demand
- •Acid test: would customers ‘scream’ if the product disappeared?
- 9:06 – 10:36
Enterprise focus with product-led growth: resolving the bottom-up vs top-down conflict
Victor explains Synthesia’s go-to-market strategy: enterprise value capture, but PLG distribution to drive adoption and generate qualified leads. He details the inherent roadmap and cultural trade-offs when serving both SMB self-serve and large-enterprise needs.
- •Decision to focus on enterprise for largest value and contract sizes
- •Distribution worked best through self-serve access and user-driven discovery
- •Bottom-up PLG and top-down enterprise sales create roadmap conflicts
- •SMB vs enterprise features often don’t overlap (e.g., admin tooling)
- •Best outcomes come from combining PLG ‘vacuum effect’ with strong sales execution
- 10:36 – 12:07
Hiring ‘bit-off underdogs’: building a hungry team over big-tech pedigrees
He argues that early-stage startups often mis-hire by chasing famous-company resumes and mismatched expectations. Instead, Synthesia looked for unconventional, hungry builders—people with grit and signals like open-source work—who could outperform more comfortable incumbents.
- •Common founder mistake: defaulting to hiring from Google/Meta/other hot firms
- •Mismatch risk: compensation expectations and big-company habits
- •Better approach: find non-obvious candidates with hustle and grit
- •Alternative signals: open-source leadership, demonstrated initiative
- •A ‘ragtag underdog’ team can outperform pedigree-heavy teams in startup conditions
- 12:07 – 13:57
Founder advice and the future of communication: from text to video/audio
Victor closes with career advice—starting a business is the best teacher and failure is usually less catastrophic than feared. He predicts a long-term shift away from text toward richer media like video, audio, and eventually AR/VR for learning and information sharing.
- •Entrepreneurship can’t be learned fully in consulting or big-company roles
- •Start sooner if you’re a builder; fear of “too early” is often overblown
- •Most employers value attempted entrepreneurship even if it fails
- •Prediction: less text consumption; more video/audio for training and knowledge
- •Long-run trajectory: increasingly immersive media (AR/VR) alongside persistent niches for text
