The Greg(g)s Say Goodbye to Claude

Why are two long-time Claude users cancelling their subscriptions? In this episode of **Two Gs and Wine**, Greg Brown and Gregg Bayes-Brown talk about why Anthropic's Claude has gone from being one of their most-used AI tools to something they increasingly struggle to work with. It isn't simply about benchmarks. Claude has become more expensive to use, increasingly verbose and, for us, harder to steer. More importantly, its tone and behaviour have started getting in the way of the work itself. So where are we going instead? We discuss: • Why we're both moving away from Claude • Token burn, unnecessary tool calls and AI assistants going beyond their brief • “Claude speak” and why AI personality actually matters • Why Gregg's trust in Claude finally broke • Greg's move towards OpenAI, ChatGPT and Codex • Why Gemini is becoming increasingly difficult to ignore • Google's growing AI ecosystem and the importance of making AI enjoyable to use • Running open-weight models locally with tools such as Goose • AI-powered cybersecurity and the rapidly changing balance between attackers and defenders • Why Chinese models are becoming serious competitors to US frontier labs • Whether the West is thinking about AI too much as a product and not enough as a capability • The hardware problem with running powerful models locally • Which AI ecosystem we think is best positioned for the next major leap • Why Google DeepMind and world models could ultimately matter more than today's LLM benchmark race The bigger lesson is that loyalty to an AI company makes very little sense. Models change. Products change. Pricing changes. Behaviour changes. The useful skill is being able to move between them and choose the tool that actually works for what you're trying to do. *Two Gs and Wine* Greg Brown & Gregg Bayes-Brown #AI #Claude #Anthropic #ChatGPT #OpenAI #Gemini #GoogleDeepMind #ArtificialIntelligence #LLM #GenerativeAI