Ceramic AI founder Anna Patterson explains their pivot from training to search was driven by a key insight: providing models with live data via low-cost search is far more efficient and timely than the expensive, slow process of continuous retraining.

Pre-reasoning AI models were static assets that depreciated quickly. The advent of reasoning allows models to learn from user interactions, re-establishing the classic internet flywheel: more usage generates data that improves the product, which attracts more users. This creates a powerful, compounding advantage for the leading labs.

LLMs can actually benefit sites with deep, authoritative content, even if it's not ranked #1 on Google. AI models prioritize surfacing the best answer, regardless of traditional rank, potentially increasing traffic for subject matter experts.

Perplexity's CEO, Aravind Srinivas, translated a core principle from his PhD—that every claim needs a citation—into a key product feature. By forcing AI-generated answers to reference authoritative sources, Perplexity built trust and differentiated itself from other AI models.

Unlike humans who type 2-3 words, LLMs generate long, sentence-like queries (e.g., eight words or more) to gather comprehensive context. This shift in user behavior from human to AI requires search engines to be optimized for these detailed, descriptive inputs.

Rather than competing to build a single foundation model, Perplexity's strategy is to be an 'aggregator orchestrator' that intelligently selects the best specialized model for any given task. This allows them to always offer the best performance without owning the underlying models, similar to how Kayak aggregates flights.

To manage non-deterministic AI products, Shopify created an internal tool where PMs grade AI-generated outputs. This creates a "ground truth" dataset of what "good" looks like, which is then used to fine-tune a separate LLM that acts as an automated quality judge for new features and updates.

Yahoo built its AI search engine, Scout, not by training a massive model, but by using a smaller, affordable LLM (Anthropic's Haiku) as a processing layer. The real power comes from feeding this model Yahoo's 30 years of proprietary search data and knowledge graphs.

Unlike chatbots that rely solely on their training data, Google's AI acts as a live researcher. For a single user query, the model executes a 'query fanout'—running multiple, targeted background searches to gather, synthesize, and cite fresh information from across the web in real-time.