James Ernst

GPT-5.1 represents the first real OS-level shift in marketing. It reasons, remembers, personalizes, and orchestrates work across channels. These are the five use cases that actually move revenue, reduce friction, and scale output in a way previous models couldn’t touch.

AI doesn’t scare me. What scares me is how fast it’s starting to sound like us. Tools like Google Pomelli mark a new phase of AI—one trained on identity, not just data. Here’s why that shift could empower marketers and disrupt agencies.

The real story here isn’t about Comet versus Atlas—it’s about context. The way we train these tools, the data we feed them, and the guardrails we set will define whether AI browsers become truly useful partners or just sophisticated echo chambers.

Perplexity AI’s Comet is a Chromium‑based browser released in July 2025 that embeds a large‑language‑model assistant directly into your tabs, promising AI‑native browsing with automation. 

For the first time in 25 years, the browser itself is changing its job description. It’s moving from displaying information to interpreting it, marking the start of what I call AI Ops in the browser—automation that lives where work actually happens.

AI browsers add a built‑in operator to your workflow—think with you, do the next step, and stay within guardrails. A practical guide for AI Ops, strategy, and optimization.

There’s too much happening in AI right now. Too many updates, too many releases, too many new products from every major player. Even if you live and breathe this stuff, it’s impossible to keep up.

Learn how to evaluate large language models for operations. A three-lens framework—Capability, Controllability, Cost—helps AI Ops teams choose, govern, and scale models responsibly.

Learn how to build a Service Level Agreement framework for AI Ops. The SLA Pyramid shows how ownership, timers, and consequences create reliable, scalable operations.