Google’s AI Search Shift: The Current State of SEM
Spring AI Product Updates and Agent Capabilities
AUTHOR:
Jen Taylor
In this edition: image and design products within ChatGPT and Claude, and the shift to agentic tools.
This is Q&AI, our blog series aimed at keeping you in the know on updates in the rapidly evolving world of AI. Sometimes, these will be quick updates on new developments in the field. Sometimes, they’ll be tips on tactics, features, or functionality. If you haven’t met me yet, hi: I’m Jen Taylor, Capacity’s Vice President, Strategic Growth and AI Activation.
Q: How have PLATFORM CAPABILITIES EVOLVED AS NEW PRODUCTS CONTINUE TO BE RELEASED BY THE MAJOR PLAYERS?
A: Here are some of the recent product updates worth taking note of:
OpenAI Image
OpenAI launched ChatGPT Images 2.0, and early coverage is positioning it as a major image-generation upgrade, especially for text rendering, structured visuals, infographics, multilingual outputs, and more polished campaign-style creative.
Early buzz is that this new model is competitive with or ahead of Google’s Nano Banana, the current market leader, when it comes to accuracy, layout and text.
Claude Design
Claude Design feels like the next logical step from Claude Code, but for visual and creative workflows. It lets users create and refine designs, prototypes, slides, one-pagers, marketing collateral, and campaign visuals through conversation, comments, direct edits, and brand/design-system inputs. I have not heard much about it yet.
Claude live artifacts
Before, artifacts were one-and-done. You’d generate something and start over every time you needed to update it. Now, you create something once and keep working inside it as the data and questions evolve. It’s the difference between a static report and a live dashboard.
Q: Where do you see the biggest cross-industry trends in how AI works?
A: The real story and shift is around agents. Google and OpenAI both launched new agent capabilities. This represents a shift from AI that helps you do the work, to AI that can take action and complete work on your behalf.
That said, “agentic” is still an industry term without a clear, agreed-upon definition, so what these tools are actually capable of (and whether they are truly autonomous agents) is still up for debate.
I also want to note that to fully realize the value of these capabilities, these tools need deep access to your systems, data, and workflows. Given the current state of data privacy, security, and governance, I’m not yet comfortable recommending adoption.
Until next time,
Jen
Have a question for a future edition? Submit it here!
