Why Gemini Argon and Meta Wonderful struggle to pull users from OpenAI and Anthropic
The author argues switching costs keep AI enthusiasts between OpenAI and Anthropic, and asks when Google's weak personal AI/agent execution becomes a real competitive threat despite Gemini Argon hype.
I see many such arguments and I agree but at what point do people update their priors? Everyone says models have zero switching costs but tbh I've only really seen people switching between the two frontier labs - OAI and Anthropic. Some folks using open source, some use Cursor here and there, but seems like most AI enthusiasts switch between OAI and Ant, which leads me to believe THERE IS some switching costs (at least today relative to the product suite, cost variance, etc.). How many folks have switched to Meta Wonderful or Gemini Argon from those two? Saw product release hype and then fade imo. Separately, having good platforms, distribution, and brand matter immensely. It was how GOOGL had high consumer AI share despite what was perceived to be a poor performing model. At what point does not executing on personal AI/agentic tools while having others interact with your product suite create serious competitive threat? We all know Google's core search business is under attack in an AI-first world, so in many ways, you would've expected them to be a lot faster-moving in this respect. I, like many, will never count them out. But, I'm posing the question, what makes us update our priors? We're multiple years into this POV - "well Google is still around and you can never count them out". At what point do people entertain whether someone out executes them and starts to become a serious competitive threat to making the google product suite just data stores for peoples' agents to interact with?
This post quotes another post: “Google's edge in persistent agents: distribution and data lock-in”