At Anthropic, we embrace a bottoms-up approach to creativity, allowing for more innovative product development close to the model's capabilities.

From day one, we built our own foundation models instead of just fine-tuning existing ones, which allowed us to make significant architectural decisions during the pre-training process.

The MCP product emerged from engineers recognizing the need for a standardized protocol to better interact with models, rather than a top-down directive.

It's important to question the narrative that foundation model companies will dominate the market. There are still opportunities for innovation outside of these models, and it's essential to avoid falling into a cynical mindset that stifles creativity.

YC creates urgency from day one, which helps founders stay focused and productive, contrasting with the often unstructured nature of starting a company.

Some critics dismiss Manus as merely a wrapper for existing foundational models and tools, but this perspective overlooks the reality that many successful AI products also fit this description.

Manus utilizes a technique called chain of thought injection, allowing agents to actively reflect and update their plans, enhancing stability and reasoning capabilities during complex tasks.

If I could give one piece of advice to YC founders, it would be to engage with real users early on to gather valuable feedback, rather than perfecting the product in isolation. I think through what the world will look like if hundreds of thousands or millions of people were to use this product or service, which helps me imagine the future of the product.

It's critical to have a clear vision and culture when scaling an organization; YC aims to produce the most innovation in a way that benefits everyone.