Embracing Digital Transformation, Episode #277

Pivot or Perish: A Startup’s Explosive Growth in the Age of Gen AI

Host: Dr. Darren Pulsipher
Guest: Joaquin Cuenca, CEO of Freepik
Release Date: July 10, 2025

Episode Overview

This episode dives into the high-stakes journey of Freepik, a major stock image platform, as it confronts existential disruption from generative AI. Host Darren Pulsipher and CEO Joaquin Cuenca discuss how startups can survive, adapt, and thrive amid waves of technological revolution. Joaquin shares Freepik’s pivot strategy, organizational changes, the vital role of user feedback, and their unique approach to leveraging (not building) AI to deliver enhanced value.

Key Discussion Points and Insights

1. Joaquin’s Origin Story & Bootstrapping in Spain

• Joaquin began coding at age 8, studied physics and computer science, and co-founded several startups. His first was acquired by Google, integrating geolocated photography into Google Maps/Earth.
• Freepik was bootstrapped due to Spain’s limited venture capital, relying on lean teams and iterative development.
• [01:41] Joaquin: “My first startup was acquired by Google... three years later I came back to Spain and co-created a couple of companies... they have been growing since then and we are pretty happy.”

2. Technology Abstraction Layers & Gen AI’s Impact

• Joaquin details how each technological wave removes barriers to starting new companies, from home-hosted servers to cloud, and now generative AI as an “abstraction layer over almost any position.”
• Gen AI allows startups to do more with less, supporting engineering, design, and marketing with a “decent” baseline of outputs.
• [04:13] Joaquin: “The next one is AI, which ... makes an abstraction layer over almost any position. So you can now have an engineer, ... designer, ... marketing person... you get something that is decent, okay, which is pretty important.”

3. How CEOs Use AI as a Double-Check

• Joaquin finds AI often more reliable than unknown human “experts,” using it to verify advice and speed up decision-making.
• [05:21] Joaquin: “When I start working with a new expert, typically I double check what they say with AI, and then... it builds trust on this new expert. But to me is helping as a double check on everything that I do.”
• Both agree AI is a “magic wand”—it speeds up work but occasionally makes small mistakes users must catch and fix.

4. Generative AI: Existential Disruption in the Stock Image Market

• Freepik’s core business: world’s most popular site for free stock images, massive user traffic, with a premium subscription model.
• The launch of DALL-E and subsequent generative image models posed a direct threat—users might now easily create unique images without traditional stock libraries.
• [12:33] Darren: “So you guys saw this as a huge threat to your business model.”
• Joaquin: “It was.”
• He describes an existential crisis—hundreds of employees at risk if they didn’t pivot.

5. Strategic Pivot: Focus on Strengths, Not Competing on AI Models

• Despite initial fears (“no right to win”), Joaquin leverages lessons from the PC revolution: as technology standardizes, the real value is in the layer above the commodity (here, the user interface, customer experience, and workflow on top of AI models).
• Freepik decided not to compete building foundational AI models. Instead, they:
  • Licensed open source models (e.g., Stability AI’s SD 1.5) and tailored the user experience for creators.
  • Focused on user feedback to improve interface and prompt design.
• [18:46] Darren: “You didn’t go off and say, ‘I need AI experts.’ ... You said, no, I’ll license that. That’s right.”

6. Building the AI Stack: Integrate, Innovate, Acquire

• Freepik took a multi-pronged approach:
  • Integration: Quickly adopted open-source models and improved UX.
  • Internal Research: Hired researchers for custom features (like inpainting, region-by-region editing).
  • Acquisition: Bought Magnific AI (creative image upscaling), which improved image quality by adding artistic detail, especially for AI-generated content.
• [23:19] Joaquin: “When something is really critical to your company... try a bit of everything.”

7. The Limits and Pitfalls of Gen AI for Stock Assets

• Not all users want AI images—some need consistency and reliability. Early mistakes included allowing too many AI assets into the standard stock library, which led to quality issues (e.g. “Why does this chair have six legs?”).
• Freepik corrected by filtering, clearly labeling, and user-segmenting AI-generated vs. traditional assets.
• [24:33] Joaquin: “We had to remove most of the AI generated content because it was not good enough. And that means detecting when it was generated, removing it from the platform...”

8. Organizational Agility and Decision-Making

• Freepik’s scale (~580 people) meant creeping bureaucracy; rapid adaptation required dismantling committees, empowering individuals, and streamlining authority.
• Key decisions reverted to individuals rather than groups, freeing the company for quick, course-correcting actions as they experimented with AI.
• [27:29] Darren: “You have to delegate that decision making to an individual ... Otherwise you get stuck and you can’t.”
• [27:42] Joaquin: “We were ruthless, like removing committees, removing group thinking, and coming back to very quick decisions, extremely fast executing them.”

9. Strategic Refocus: The Professional User as Target Market

• With consumer-grade generative image creation commoditized by ChatGPT/Google/etc., Freepik’s future lies with creative professionals who need detailed control, collaborative workflows, and higher-quality outputs (video, audio, specialized images).
• Strategic decisions (set at the top) are translated into tactical actions closer to where the work is done—product teams and engineers are deeply involved with user feedback and empowered to act.
• [30:53] Joaquin: “The decision high level is we work for professionals. What they need that they don't have today is control over the image...”

10. Engineers’ Role and Skillset Evolution

• Modern engineers must engage with users, understand business needs, and creatively solve problems—not just “take orders.”
• AI enables smaller, more effective product teams; tools automate mundane tasks and summarize user interviews, further tightening the feedback loop.
• [34:12] Joaquin: “If I need precision, what I want, I'm not gonna need you in one year, man... you need to understand the user, ... come up with ideas...”

Notable Quotes & Memorable Moments

• On Disruption:
  [14:51] Joaquin: “We were at the time around 500 people. ... How do we deal with this and how we can turn the company into something that is useful to people and still be relevant.”
• On Gen AI as Magic Wand:
  [07:26] Joaquin: “The metaphor of a magic wand is quite spot on because ... sometimes there is a spell that doesn’t work ... but it is overall right...”
• Pragmatic Approach to AI Build/Buy:
  [19:00] Darren: “That cost of retraining all your people and not relying on your strengths is astronomical and it will kill a company.”
• On Adaptation:
  [25:35] Darren: “You can move quickly, you can adapt to a too much AI, back off a little bit. In large companies, they have a lot harder time with that...”
• On Leadership in Growth:
  [27:42] Joaquin: “We were ruthless, like removing committees, removing group thinking, and coming back to very quick decisions, extremely fast executing them.”
• On Engineers’ Future:
  [34:12] Joaquin: “If I need precision, what I want, I'm not gonna need you in one year, man... you need to understand the user, ... come up with ideas...”

Timestamps for Important Segments

• 01:41 — Joaquin’s background and bootstrapping in Spain
• 04:13 — Generative AI as a new abstraction layer
• 08:16 — Freepik’s history and the existential crisis from AI
• 12:33 — Recognizing generative AI as disruptive threat
• 18:46 — Pivot: Focused on strengths, not building models
• 23:19 — Multi-pronged AI adoption: integration, innovation, acquisition
• 24:33 — Realizing the limits of Gen AI for core stock content
• 27:42 — Organizational changes to speed up adaptation
• 30:53 — Shifting strategy to focus on professional users
• 34:12 — The evolving role of engineers in the age of Gen AI

Conclusion

Joaquin and Darren’s conversation is a compelling roadmap for how startups can not only survive but also harness disruptive technologies. By focusing on core strengths, integrating rather than competing on AI, being agile in organization and product, and listening closely to users, Freepik was able to pivot quickly and find new growth in the age of generative AI.

For anyone navigating digital transformation—especially in high-disruption sectors—this episode is a blueprint for resilience, adaptation, and strategic focus.

Find out more about Freepik at freepik.com.
For more episodes, visit embracingdigital.org.