Insights · Analysis
Analysis of how ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok, DeepSeek, Meta AI, Google AI Overviews, and AI Mode are rewriting discovery, and what to do about it.
91 insights · search & filter
OpenAI's new Computer History feature lets ChatGPT remember what you do across every app and website on your machine. Personalisation just moved from what people type to what they actually do all day, and it changes how brands get surfaced before a query is ever asked.
AI search is not one channel. It is three architectures that reward different work, and the platform that cites you most refers almost no one. Here is the taxonomy, the numbers, and six myths that keep defunding the channel a year too early.
New analysis of a few hundred ChatGPT search queries shows the model names brands from memory before it fetches a single page. Being in that query is worth around 33 times more than being findable. Here is what it means for your GEO strategy.
Google Ads is previewing Ask Advisor, an AI assistant that pulls context across platforms to shape creative and strategy. For growth leaders this is a shift from managing campaigns to delegating them. Here is how to prepare so the agent advises well.
The EU has fixed 2 August 2026 as the date AI transparency rules bite. From that point, AI content needs machine-readable marks and people must be told when they are dealing with a machine. Here is what marketers actually need to do.
Meta is reportedly scraping the web at scale to build its own search index, one it can feed to its AI without leaning on Google or Bing. Here is what happened, why it matters, and the concrete steps growth teams should take on crawler access now.
OpenAI's new Signals dataset shows ChatGPT users are twice as likely to complete a task at work as to ask a question. For growth and search leaders, that is the clearest primary-source proof yet that commercial intent now lives inside the assistant. Here is how to read it as a discovery-channel map.
OpenAI has classified its upcoming Astra model as potentially 'critical' for cyber capabilities, gating it behind hard safety controls. That gating instinct is spreading into the discovery layer, where verifiable, authoritative content wins AI citations and thin content gets filtered out.
Claude Code now auto-approves commands by default, handing AI agents real autonomy. The safety net is a classifier that caught 89% of dangerous actions against 13.6% for human review. For marketing teams running agents in production, this is a governance decision, not a convenience feature.