AI Newsletter

Anthropic has overtaken OpenAI for the first time in B2B customer usage (34.4% vs. 32.3%), driven by explosive growth over one year. The lead signals a market shift in AI competition, but remains fragile given intense competitive pressure and rapid product development.

Meta is introducing an Incognito mode for its AI assistants where chats are not stored and Meta itself has no access – a strategic move to differentiate from competitors like OpenAI and Google, which use user data for training. This addresses growing data privacy concerns and could increase trust in Meta AI, while enabling sensitive use cases (health, finance).

Anthropic is positioning Claude as a vertical solution for SMB core processes (accounting, invoicing, marketing) with pre-integrated workflows and training. With this, the company is addressing a highly profitable mass market that no AI solution has served with this specificity so far, and creates strong switching costs through ecosystem integration.

Luma is positioning itself with Uni-1.1 as a strong third player in the image generation API competition and undercuts established competitors through aggressive pricing (0.04 $/image). The model closes the quality gap to Google and OpenAI, while integrated web search and enhanced reference image features provide additional differentiation and could incentivize developers to switch platforms.

DeepMind is developing "Pointer Engineering" – a new AI interaction method that complements or replaces prompt engineering – instead of text, users provide visual pointers/coordinates. This could simplify the usability of AI systems and democratize access to complex applications, especially for non-technical users.