Start a project

AI this week: agents get their own app stores

Our pick of the week’s AI news, and what each story means for people designing websites and products. Agents, cheaper models and AI search.

On this page
  1. Chat assistants become app platforms
  2. Cheaper, faster models
  3. Agents at the checkout
  4. Natural-language search spreads
  5. Keeping agents in check
  6. Names that change how people think

Key takeaways

  • Chat assistants are turning into platforms with their own apps and interfaces.
  • Faster, cheaper models make AI features practical in more everyday places.
  • Agents are starting to use websites directly, so structure and clarity matter more.

Each week we pick a handful of AI stories that matter for people who design and build websites, and add a short note on why. The news comes from the publications linked; the “why it matters” notes are our view. For the full stream of headlines, see our AI news page.

Chat assistants become app platforms#

At its developer conference, OpenAI announced ways for ChatGPT plug-ins to show app-like interfaces and run automations, and TechCrunch described the new features as a direct challenge to the app store model. It also launched “Dots”, an agent with an animated avatar. TechCrunch

Why it matters: more services will be used inside an assistant rather than on their own website. Designers will need to think about how their product looks and behaves in someone else’s interface, in a few small cards rather than a whole page.

Cheaper, faster models#

Anthropic released Claude Sonnet 5.5, which it describes as a significantly cheaper and faster work partner. OpenAI launched GPT-6.1 Sol, which it says nearly matches its larger model at a lower cost. TechCrunch · TechCrunch

Why it matters: speed changes what AI features feel like. Faster responses bring more features under the point where people stay engaged, and lower costs make AI practical in places that couldn’t justify it before, such as search boxes and forms.

Agents at the checkout#

Shopify opened its checkout to browser-based AI agents, according to TechCrunch. TechCrunch

Why it matters: some of your “visitors” may soon be software acting for a person. Clear structure, predictable forms and honest labels help both humans and agents, which is one more reason to fix confusing checkouts now.

Natural-language search spreads#

Airbnb added AI-powered search alongside new social features. TechCrunch

Why it matters: people are getting used to describing what they want in their own words. The hard part is the first move: an open search box needs good suggestions to get people started, which we covered in The empty prompt box problem.

Keeping agents in check#

Nvidia launched a platform aimed at reining in rogue AI agents, and TechCrunch reported on why OpenAI is absent from that industry-wide effort. TechCrunch

Why it matters: as agents do more on people’s behalf, oversight moves from a technical concern to a design one. People need to see what an agent plans to do and approve the important steps; see Checkpoints, not blind trust.

Names that change how people think#

Google is replacing Gemini’s “Gems” with “skills”, according to TechCrunch. TechCrunch

Why it matters: a name sets expectations. “Skills” suggests something an assistant can do, not a separate thing you create. When you name features, choose the words that match what people will actually do with them.

Share this article

Keep reading

Related articles

All articles

Newsletter

New writing, once a month.

Practical notes on design, search and research. No spam, unsubscribe anytime.