Product Announcements

AI tools can now (easily) talk to your other apps

This past week brought two major announcements that solve a problem you've likely been thinking about: how to get AI to work with the tools you already use. To that end, Anthropic launched Integrations for Claude, and OpenAI rolled out Connectors for ChatGPT. Both let you connect directly to a wide range of productivity and file storage applications. Imagine, for example, asking Claude to find that design brief buried in Google Drive, or having ChatGPT pull project updates from Slack.

What makes this different from previous "AI integrations" is the simplicity. Both are built on the same foundation: the MCP (Model Context Protocol), the technical standard that enables these connections. Instead of complex API setups or clunky workarounds, both companies are offering plug-and-play connections.

The simultaneous launch of OpenAI and Anthropic's solutions demonstrates how this type of integration is becoming a table-stakes requirement for AI assistants.

Claude can now connect to your world
⚡ Quick Read (2 minutes) - exclusive of embedded demo videos

Is this the biggest ChatGPT update for business users?
Watch Time: 8 minutes

Product Announcements

Apple opens its AI to third-party apps (but Siri still waits)

Since unveiling Apple Intelligence at last year's (2024) Worldwide Developer Conference (WWDC), Apple has struggled to deliver on its promise for AI on its devices. Before this year's WWDC, there was speculation about how much Apple would focus on AI in this year's software releases. Most significantly for developers, Apple announced the Foundation Models Framework--a way for third-party apps to tap into Apple Intelligence. Because Apple Intelligence operates on-device, this API enables developers to implement AI in a manner that is faster, less expensive, and more private than cloud-based AI solutions.

For consumers, the Apple Intelligence announcements were modest. The long-awaited AI overhaul of Siri, once billed as a cornerstone of Apple's Intelligence, remains absent. But there were new features announced. Workout fans might enjoy the new voice-enabled "Workout Buddy" in the Fitness app, and they showed an upgraded, AI-driven Spotlight for macOS that could give apps like Raycast a run for its money. They also showcased live translation for Messages, FaceTime, and Phone. Mashable has a complete rundown on all the Apple Intelligence announcements:

WWDC 2025: All the Apple Intelligence AI features coming to your devices
⚡ Quick Read (3 minutes)

AI in the Organization

The end of enterprise systems as we know them

I made my career designing enterprise workflow applications. All those step-by-step forms and other highly structured screens? They cease to exist in what HBR calls AI-enabled systems of work:

"In a CRM context, an AI-driven system would analyze signals—emails, call transcripts, proposal revisions—and automatically update a lead’s status. It would learn from historical outcomes to refine the underlying process, eliminating manual data entry and reducing process latency.

Likewise, complex queries—once the domain of data analysts translating requests into structured query language (SQL)—become a matter of simple conversational prompts: “How many qualified leads converted to deals exceeding $100,000 in Germany this quarter?” Generative AI systems can parse unstructured inputs, generate the necessary database queries, and deliver answers in seconds."

This article is an excellent primer on the impact of AI on enterprise systems.

How Gen AI could disrupt SaaS
☕ Medium Read (6 minutes)

AI in the Organization

Why aren’t organizations realizing productivity gains with AI?

"AI use is ubiquitous and leads to performance gains at the individual level that are not passed on to organizations." - Ethan Mollick

Data are increasingly showing that while individual workers are quietly saving hours with AI tools, most aren't telling their bosses about it. People may fear their AI use will be punished, they may be getting rewarded for outputs they have (secretly) prepared with the help of AI, or they may worry about job cuts if the company discovers that AI is handling some of their work. Ethan Mollick refers to people keeping their AI use hidden as "Secret Cyborgs."

The real challenge for organizations isn't the technology; it's creating environments where people feel safe experimenting and sharing what actually works. Ethan Mollick presents a three-part organizational framework for making this happen: Leadership, Lab, and Crowd.

Making AI Work: Leadership, Lab, and Crowd
⚡ Quick Read (4 minutes)

AI for Research

Getting human-quality analysis of user interview transcripts

Dr. Nick Fine shares the prompt he uses to get what he considers to be "human quality output" from Gemini 2.5 Pro when analyzing user interview transcripts. His experience with this prompt is another example of how the latest frontier models are providing much more reliable time-saving value.

Of course, there is a big caveat here: poor interview practices can lead to biased insights. In other words, garbage in, garbage out. But in the right hands (an experienced interviewer), Nick Fine’s approach feels like a game-changer.

Using Gemini 2.5 Pro to Analyze User Interview Transcripts
⚡ Quick Read (2 minutes)

That’s it for this week.

Thanks for reading, and see you next Wednesday with more curated AI/UX news and insights. 👋

All the best, Heidi