Lessons from a fireside chat with Perplexity's CEO

"The highest dopamine rush wasn't from raising millions."

Aravind, Perplexity's CEO, smiled as he shared this at our fireside chat.

"It was reading the first 3,000 queries users made to our engine."

That hit home.

Because real impact matters more than headlines.

Here's what else he shared:

• Application layer is where breakthroughs happen, not the companies making headlines with foundation models.

• The economics are transforming model capabilities double while costs halve every 6 months.

• AI agents are the next frontier, moving beyond chat to complete end-to-end tasks autonomously.

• Search is evolving beyond 10-blue-links to direct answers and actions.

• Even smaller models (like Llama3.1-8B) are outperforming larger previous-gen models.

• Silicon Valley is shifting focus from building foundation models to creating powerful applications.

The next 12-18 months will be transformative.

As inference costs plummet and models become more capable, we'll see AI agents handling end-to-end workflows across industries.

The winners won't be those with the biggest models, but those who build the most impactful applications.