The Twenty Minute VCRoundtable #7: Spotify, Adobe and Linkedin on How AI Changes The Future of Product & Design | E1097
At a glance
WHAT IT’S REALLY ABOUT
Spotify, Adobe, LinkedIn Reveal How AI Redefines Product, Design, Strategy
- Product leaders from Adobe, Spotify, and LinkedIn discuss how AI is shifting product development from UI-centric to AI-centric, where AI becomes the product and UI becomes the signal-capturing interface around it.
- They argue that AI-first strategy must come from the CEO down, forcing teams to accept probabilistic, non-deterministic experiences, rethink design, and deeply understand models and data rather than delegating AI to a specialist team.
- The conversation explores multi-model architectures, routing and cost optimization, proprietary vs. foundation models, and why owning high-quality, proprietary user data is a durable advantage.
- They also examine how AI will disrupt business models (e.g., seat-based SaaS and hourly billing), what skills designers and product managers must build, and why incumbents may be unusually well-positioned in this particular platform shift.
IDEAS WORTH REMEMBERING
5 ideasTreat AI as the core product, not a feature you delegate.
The leaders argue that AI strategy must be driven from the top and embedded in every product decision; product teams can’t outsource AI thinking to a separate “AI team” if they want to remain relevant.
Designers and PMs must understand models as deeply as users.
Effective AI UX requires knowing capabilities, failure modes, and latency/cost tradeoffs of models (e.g., Midjourney’s tolerant UI design), so teams can shape experiences that match current model performance.
AI shifts experiences from deterministic to probabilistic, demanding new mindsets.
Product leaders must accept they no longer fully control outputs; they set objectives, constraints, and “knobs,” while models generate variable outcomes—sometimes “hallucination” is a feature (e.g., creative tools, discovery).
Multi-model architectures and routing will be a major innovation layer.
Instead of a few mega-models, they foresee many specialized models (some local, some open-source, some proprietary) orchestrated by dispatchers/routers that optimize for relevance and cost, abstracted away from most product teams.
High-quality, proprietary user data and objectives design are critical advantages.
They stress that data is “oxygen” for AI: product leaders should own the definition of algorithm objectives and the design of data collection pipelines, rather than assuming data science will fix poor or missing signals later.
WORDS WORTH SAVING
5 quotesAI is the product and the UI is there to help the AI.
— Gustav Söderström (Spotify)
When you're thinking in an AI-first principled kind of way, you're really unleashing the idea of control. What happens basically in AI is you don't control the experience anymore.
— Tomer Cohen (LinkedIn)
Designers and product people need to understand GPT-4 as well as they understand a user.
— Gustav Söderström (Spotify)
It's not the technology that makes us successful, it's the user's experience of the technology that makes us successful.
— Scott Belsky (Adobe)
Data is literally, according to me, your second most important job. The first is understanding the objective of the algorithm.
— Tomer Cohen (LinkedIn)
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome