Silicon Valley Girl$18B AI CEO: How to Build a Million-Dollar Business in the Age of AI
CHAPTERS
- 0:00 – 1:15
AI flips the playbook: speed, uncertainty, and why 12-month plans break
Marina sets up Andrey’s core thesis: AI changed software creation so dramatically that product cycles, competition, and predictability look different now. Andrey argues that in this environment, founders must move fast, stay mission-driven, and accept that long-range forecasting is unreliable.
- •AI is the major change Miro didn’t predict
- •Planning beyond ~12 months is increasingly unrealistic
- •Software is easier to build, but the market is more crowded
- •Founders need speed plus clarity about what they’re passionate about
- 1:15 – 3:42
From agency pain to browser whiteboard: building without a grand ambition
Andrey explains Miro’s origin story: a practical need for remote collaboration while running a creative agency. Early goals were modest—solve a real problem and reach break-even—rather than aiming for a giant valuation or user count.
- •Idea came from collaborating with remote clients
- •Initial goal was break-even, not “build a unicorn”
- •Early product didn’t work; the team rebuilt based on signals
- •Focus on solving the problem best-in-class and choosing a market with upside
- 3:42 – 5:50
Growth inflection points: Flash→HTML, enterprise motion, and the pandemic surge
Andrey walks through the “aha” milestones where Miro’s potential became clearer—first technical platform shifts, then enterprise sales capability, then explosive adoption during the pandemic. He also describes post-pandemic flattening and the continued path to 100M users.
- •2015 shift from Flash to HTML clarified a path to meaningful revenue
- •2018–2019: enterprise sales unlocked a bigger growth curve
- •Pandemic: 5M to 50M users in ~18 months; later growth stabilized
- •Miro surpassed 100M users with ongoing platform expansion
- 5:50 – 6:51
How Miro gained users fast: virality, delight, and SEO before “real marketing”
The conversation turns to distribution: early Miro prioritized product experience and built a natural invitation loop. After the organic flywheel worked, the company layered on more intentional marketing and sales.
- •Optimized onboarding and UX for a delightful first experience
- •Designed collaboration features that incentivize inviting others
- •Word-of-mouth and virality as core early growth engines
- •SEO became a major channel; marketing/sales scaled later
- 6:51 – 7:41
Product-market fit in 2025: fundamentals stay, brand and trust matter more
Andrey argues that even if building is cheaper and faster with AI, PMF still determines whether a product grows. He emphasizes that perceived quality, brand recognition, and trust become differentiators when products are easy to replicate.
- •AI makes building faster, but quality still requires investment
- •PMF remains the non-negotiable foundation for growth
- •Brand, trust, and “lovemark” energy are stronger moats now
- •Crowded markets punish products that don’t solve a real problem
- 7:41 – 9:03
Finding PMF: problem framing, customer interviews, and prototyping for signals
Andrey outlines a practical approach to discovering PMF: validate the problem, understand market size, and talk to customers with open-ended questions. He notes that especially for AI-first productivity tools, customers may not articulate what they need—so prototypes are essential.
- •Start with: real problem + big enough market
- •Use open-ended customer conversations to test hypotheses
- •In AI-first UX, customers may not specify needs upfront
- •Prototype early; 7–20 deep interviews can reveal strong signal
- 9:03 – 10:58
Why you need to fail: targeting a 30% failure rate and iterating on solutions
Marina and Andrey discuss experimentation discipline: if everything works, you’re not pushing enough. Andrey explains how to manage a portfolio of bets (safe + moonshots) and how to decide whether to pivot the problem or iterate on the solution.
- •Healthy experimentation: ~50–70% success, ~30% failure
- •Applies to product/growth experiments and even acquisitions
- •Balance predictable bets with moonshots that may fail
- •Diagnose: wrong problem vs. imperfect solution; iterate when problem is real
- 10:58 – 12:20
Redesigning for AI: “day-one thinking,” onboarding rewrites, and AI Canvas as a bet
Andrey describes Miro’s continuous iteration mindset—hundreds of onboarding redesigns over 14 years—and how AI prompted a new interface experiment. He explains AI Canvas as a testable mode that may merge with or remain separate from the core experience depending on user behavior.
- •Iteration never stops; user expectations and UI possibilities change
- •AI Canvas introduced as a new interface and capability set
- •Positioning is uncertain: separate mode vs. merged core
- •Launch, observe adoption, learn, and be willing to kill/merge
- 12:20 – 14:21
Trust as the new currency (Sponsor): privacy expectations and Cookiebot CMP
Marina emphasizes how trust and privacy concerns shape product adoption, especially with AI apps handling sensitive data. She shares stats on consumer privacy anxiety and introduces Cookiebot CMP as a consent-management solution to help businesses comply with regulations.
- •Users worry about where their data goes in AI tools
- •Privacy concern is widespread (statistics cited)
- •Risk: unknowingly breaking privacy laws on websites
- •Cookiebot CMP: scanning, consent management, regulation updates, integrations
- 14:21 – 15:38
Standing out in commoditized software: name vs brand vs “lovemark” (RealtimeBoard→Miro)
Andrey explains Miro’s rebrand logic as an antidote to sameness in software. He distinguishes descriptive names from brands and from “lovemarks”—names that evoke emotion—and connects “Miro” to the artist Joan Miró and the idea of an inspiring canvas.
- •Need to stand out as software creation becomes commoditized
- •Three layers: descriptive name → brand → lovemark
- •Goal of rebrand: move from literal utility to emotional connection
- •Miro name inspired by Joan Miró and the “canvas” metaphor
- 15:38 – 20:42
Strategy resets in the AI era: three-year “paint the picture” vs six-month commitments
Andrey shares Miro’s historical practice of three-year vision documents and why it’s being reconsidered. With models and ecosystems changing quickly, he argues for shorter planning horizons (six-month commitments) while keeping a stable mission as the North Star.
- •“Paint the picture” (from Atlassian) used for multi-year alignment
- •AI disrupted prior long-range assumptions
- •Now: commit to ~6 months; keep a looser 12–18 month view
- •Focus on mission and where you have “permission to win” in crowded markets
- 20:42 – 24:48
Why building is more exciting now + AI Canvas demo: multimodal flows and sidekick editing
Andrey explains why LLMs expand the solution space and make product building feel like a ‘candy shop’ of possibilities. He then demos AI Canvas—connecting references, prompting image generation, choosing models, and iterating via a conversational ‘sidekick.’
- •LLMs enable multiple new ways to solve the same problem
- •Software interfaces and “surfaces” are being reinvented
- •Demo: connect reference inputs → prompt → choose model (e.g., Stable Diffusion)
- •Iterate outputs with follow-ups (variants, object removal) and multimodal steps (tables/Kanban)
- 24:48 – 35:02
Founder edge and where to build: curiosity, vertical AI, favorite tools, and the daily question
Andrey closes with founder advice: continuous curiosity, critical thinking, and resilience amid accelerating change. He highlights “founder-market fit,” predicts vertical AI will scale fast (legal, coding, marketing), shares books and AI apps he uses, and ends with a simple daily self-check for motivation.
- •Most important traits: curiosity + critical thinking + resilience
- •Founder-market fit: build where your strengths and passion align
- •Vertical AI opportunities: legal, coding (shift to context engineering), marketing/content and ROI optimization
- •Recommendations: High Growth Handbook, High Output Management; apps like Granola plus major LLM assistants
- •Daily question: “Do I love what I’m doing?”
