Apple Liquid Glass: UX genius or accessibility nightmare?
Plus: How to build an effective AI product strategy, WhatsApp gets AI summaries but Wikipedia ditches theirs.
Hi product people , Rich here with your weekly briefing and a warm welcome to the 234+ new subscribers who joined us since last week. The Department of Product is now read across all 50 US states and 160 countries worldwide - thank you for your support 🙏.
Coming up this week, a closer look at Apple’s Liquid Glass and the future of mobile UX, why Wikipedia has ditched AI summaries (for now) and a new AI agent startup that’s designed to make building internal apps that are aligned with your design systems super simple.
Happy Friday and enjoy the weekend ahead!
More: Watch on YouTube / Follow on Substack Notes / Unlock the Full Newsletter
Liquid Glass and the future of mobile UX
First up, Apple’s WWDC kicked off this week and compared to the other recent events from Google, Microsoft (and even Dyson!), this one felt strangely flat. There was an undercurrent of “don’t mention AI” and some of the announcements that were made didn’t quite hit the mark. On the plus side, iPad OS finally seems to have nailed multi-tasking and Apple Wallet got some neat Digital ID support updates.
But for product teams, of course, the most significant announcement is the all-new design language that Apple is calling Liquid Glass. Many companies have tried and failed to do this, the most notable being Microsoft’s Windows Vista. To be fair to Apple, many of the issues with Vista glass effects UI were related to processor limitations that Apple users in 2025 are unlikely to run into.
The main concern though, is accessibility. And because of this, the reaction to Liquid Glass has been mixed to say the least. But if you are curious about what a Liquid Glass-ified version of your own product might look like, Apple’s comprehensive design guidelines are worth a read. Some of the most important of these guidelines include:
Moderation in Design - avoid overusing Liquid Glass to maintain content focus. Apple says that designers have noted readability issues with excessive translucency, so prioritize clarity. Plenty of social media posts make this point pretty clear.
Test with Accessibility Settings - ensure interfaces adapt to settings like “Reduce Transparency” or “Reduce Motion.” Standard components handle these automatically, Apple says, but custom elements must provide fallback experiences (e.g., reduced animations or opaque backgrounds).
Minimize Custom Backgrounds - avoid custom backgrounds in controls, navigation elements, or views (e.g., split views, tab bars, toolbars) to prevent interference with Liquid Glass’s translucency and scroll edge effects. Let the system handle background appearances.
Meanwhile…
The Browser Company has officially released Dia in a beta for existing Arc users, the agentic browser set to replace Arc. I’ve been playing around with it for the past few days and so far, so good. It’s still a little bit buggy since it’s only a beta version but I’ve found it to be super helpful for things like chatting with PDF reports or summarizing long articles. The UX strays far enough from Chrome to make it feel different but not so far as Arc did to make it feel like a challenge to switch. See some of these use cases in action here.
Google’s Gemini has released Scheduled Actions on its mobile app - a new feature that allows you to ask Gemini to perform a task at a specific time. For example, you can ask it to provide a summary of your upcoming meetings or unread emails every morning. Gemini has also been added to Google Forms which lets you summarize responses and Google Sheets now lets you edit charts created with generative AI.
Google’s VP of Gemini Josh Woodward also unveiled big updates to its Veo3 video generation which lets users generate videos 2x faster.
A new AI Agent called Clark has been released by AI startup Superblocks. Clark is designed for product teams to build internal apps quickly. Users simply describe the type of internal apps they want to build and Clark handles the rest.
Is Meta falling behind?
Meta seems to be falling behind in the AI arms race and this week announced $15bn in funding into AI data labelling startup Scale AI to build an AI “super intelligence” team. Meanwhile, its users seem to be quite confused about how the Meta AI public feed works. There have been several examples of users posting personal, private information on Meta AI’s public feed which shows that some users are very confused about the prospect of a public feed integrated into these tools.
WhatsApp is getting AI-powered summaries on mobile. New leaks show that once activated, if there’s a large enough message backlog, a new “Summarize with Meta AI” button appears in place of the usual “X unread messages” label. Tapping it returns a brief recap of the conversation.
Deepmind’s Google CEO on the future of work
Key reads and resources for product teams
New from the Department of Product Substack this week:
Knowledge Series - How to become “AI fluent”
Tech CEOs are demanding AI fluency. Here’s some practical ways to demonstrate it. In this Knowledge Series, we take a closer look at some of the practical ways product teams can demonstrate AI fluency in 2025.
Deep dive - How companies are using AI internally: use cases from Anthropic, Uber, Netflix, OpenAI and more
Feature evaluation, CRM enrichment, legal changes, tech debt management, data analysis, personalised PRDs. Internal AI use cases you can use as inspiration for your own product. (Department of Product)
More reads and resources:
How to build an effective AI strategy (Google’s Head of AI Solutions)
How Expedia is preparing for a future where people use AI to plan their vacations (Business Insider)
Lyft’s CEO on customer-obsessed innovation (Harvard Business Review)
Tools you can use
Helium - a new startup that lets you A/B test paywalls and improve conversions. Their co-founder says its ultimate goal is to replace the product manager.
Chronicle - a new presentation tool that uses pre-built widgets to make it easier to build beautiful presentations.
Exa - business grade search for competitor research, market analysis and more.
📈 Product data and trends to stay informed
Apple mentioned Siri just twice in this year’s WWDC keynote:
YouTube says its ecosystem created 490,000 jobs and added $55 billion to the US GDP in 2024. The figures were released as part of its annual impact report.
OpenAI has hit $10 billion in annual recurring revenue and is projected to hit $125 billion by 2029.
Amazon’s employees have said that AI is making them start to feel like warehouse workers. But GitLab’s CEO told investors that he expects the number of software engineers to grow thanks to AI. “I think the number of people who are able to create code is only going to increase with some of the power of AI and ability to create code without necessarily the deep technical skills at times required.” he said on an earnings call. Read the transcript here.
99% of hiring managers are now using AI in some capacity during the hiring process and 53% of hiring managers say they can tell when candidates use AI for applications. Report on global hiring trends.
New analysis shows that ChatGPT will prioritize saving itself over humans with diabetes and other conditions. In simulated tests, ChatGPT chose self-preservation over user safety up to 87% of the time in some scenarios.
SaaS customers are struggling to understand outcome based pricing. 39% of respondents say they prefer usage based pricing with lack of clear outcomes and unpredictable costs as the main reasons for this. AI enterprise report 2025.
Finally…
The “Like” button is something we all take for granted in 2025. But this is thought to be the first ever conceptual sketch of a “Like” button, hand drawn by Yelp’s Bob Goodman:
Paid subscribers get the full DoP Substack including: The Knowledge Series for sharpening their tech skills, AI tutorials for putting AI into practice at work and DoP Deep dive reports to learn from the world’s top tech companies.