ChatGPT Images 2.0, Claude 4.7, and the Great Divergence in AI adoption

OpenAI takes the image-gen crown. Anthropic's Opus 4.7 lands with vision improvements. And new research shows the top 20% of companies are capturing 75% of AI's economic gains.

A model-release-heavy week with the bigger story being the widening AI capability gap between leading companies and the rest.

OpenAI releases ChatGPT Images 2.0

My take: A step change in image generation: superior detail, much better instruction-following, dense and accurate text rendering, real-time web search for current information. The interesting workflow is GPT Image 2 to Codex: generate a UI mockup, then turn it into working code. If you do any product or marketing work, this collapses a lot of design-to-production time.

Anthropic releases Opus 4.7

My take: Incremental but significant. Better at longer end-to-end reasoning tasks (less babysitting), improved vision for whiteboards, dashboards, and charts. The key implication: you can now hand off bigger chunks of work to Claude with less hand-holding. Worth revisiting any workflow where you previously had to break tasks into small steps.

Claude Design suite launches

My take: A new UI generator that creates 'systems design' (websites, apps) for non-designers and Claude Code users. Uses code and SVGs rather than image generation. Limits are tight, but the direction matters: designing and building software is converging into one workflow. Plan how your design and engineering teams will work together when the line between them gets even blurrier.

Codex becomes a general knowledge-work platform

My take: Computer Use (Mac only) lets Codex click and type across any application. Heartbeats let it run scheduled recurring tasks in the same thread, maintaining context over time. This is the 'monothread' or 'chief of staff' pattern. If you want one practical experiment to try this quarter, set up a long-running thread with Codex acting as your operations assistant.

PwC: top 20% of companies capture 75% of AI's economic gains

My take: Leaders use AI for growth and business model reinvention. Laggards use it for efficiency. The compounding implication is brutal: the gap between leaders and laggards isn't closing, it's widening. If you're at a leader company, push harder. If you're at a laggard, the time to fix it is now, not 2027.

Stanford AI Index: expert optimism vs public pessimism

My take: A wide gap between how AI experts and the general public view AI's impact. Worth thinking about as a leader. Your customers and employees probably skew toward the public view, not the expert view. Adoption strategies that ignore this end up looking tone-deaf or even threatening. Communication matters as much as capability.