“In technology, whoever controls the platform controls the narrative,” as several strategic analysts have observed. The rise of AI is testing that maxim in new ways. A single large language model can be both the underlying platform that developers build on and the end-user product millions adopt directly. For companies in the AI era, the question is no longer whether to be a platform or a product, but how to navigate being both at once.

The Blurred Line in Practice

Consider OpenAI. The company provides an API that powers thousands of applications, making it a platform. At the same time, it operates ChatGPT, one of the most widely used consumer products in the world, built on that very same infrastructure. Anthropic follows a similar pattern, offering Claude as a developer-facing API while also positioning Claude Code as an integrated product experience for knowledge workers.

These examples highlight the duality at the heart of AI strategy. Platforms attract developers and extend reach. Products capture direct users and create faster feedback loops. AI companies are increasingly straddling both roles out of necessity.

Commoditization Pressure

The urgency comes from commoditization. Core LLMs are now accessible from multiple providers, including OpenAI, Anthropic, Cohere, and open-source projects such as Meta’s LLaMA. When the underlying models are interchangeable, differentiation shifts elsewhere. Companies must either:

  • Own the product experience, turning the model into a daily workflow or consumer habit.
  • Own the platform ecosystem, building a stickier developer environment of integrations, tooling, and distribution channels.

The danger lies in being stuck between the two—neither a beloved product nor a thriving platform, but a utility with no defensible edge.

Historical Parallels

This tension is not new. Microsoft built Windows as a platform, but also created Office as a product to drive adoption and revenue. Apple took the same route, pairing iOS with a suite of native apps to showcase the experience. In both cases, the platform and product reinforced each other.

What’s different in AI is the cycle time. With models updating every few months and user adoption moving at internet speed, companies must navigate platform–product strategy in real time. Decisions that once took years in the PC or mobile eras now compress into quarters.

Future Speculation

One plausible future is the rise of platform-products—hybrids where the line between app and API vanishes. ChatGPT plugins already move in this direction, turning a consumer-facing product into a platform for third-party developers. Claude and Perplexity are experimenting with integrations that extend their utility beyond the core chat interface.

This suggests a future where every AI product is also a developer surface, and every platform doubles as an end-user tool. The analogy might be an “AI-native app store,” but one that lives inside the product itself rather than as a separate layer.

Takeaways for Product Managers

For product managers and technologists, three lessons stand out:

  1. Positioning matters. Clarity on whether you are targeting builders or end users is critical, even if you serve both.

  2. Feedback loops create moats. Products generate user data that strengthens the platform layer. Platforms enable broader adoption that can feed product improvements.

  3. Convergence is the default. In AI, expect most companies to operate simultaneously as platforms and products. The winners will balance these roles without diluting either.

Conclusion

AI is erasing the boundary between platforms and products. The same model can be an API, an app, or both, depending on context. Historical playbooks offer clues, but the pace of change is faster and the stakes higher. The companies that succeed will be those that embrace convergence, creating ecosystems and experiences that reinforce each other.