🔵 Stripe's new homepage, Chrome gets agentic powers and a warning from Anthropic's CEO

Plus: A new Superagent from Airtable, How to create high density McKinsey-grade presentation decks and SaaS stocks are in trouble

Hi product people 👋,

Welcome back. We’re officially one month into 2026 and I hope your year’s off to a good start so far.

This week, Anthropic’s CEO published a 20,000 word piece on his predictions of where we’re headed over the next 1-5 years. We’ll dig deeper into that piece along with some real world examples of how his predictions might already be impacting product teams.

Plus, a look at Stripe’s new homepage, new benchmarks showing even the best AI agents still fail 75% of complex workplace tasks, and OpenAI’s scramble to monetize with $60 CPMs and revenue cuts from scientific discoveries.

If you have any views on anything in this week’s Briefing, just drop a note below!

Rich

Watch on YouTube | Follow on Substack Notes

---

Key reads and resources for product teams

Deep - AI voice in Practice: How product teams are using AI voice features

The latest DoP Deep dive explores how leading companies are using AI voice agents to transform core product workflows - from user research and onboarding to international expansion and product demos. Drawing on real-world examples from ElevenLabs, Todoist, Harvey, YouTube, Shopify and more. You’ll come away with practical design principles, strategic questions to decide if voice makes sense for your product, and ideas for turning AI voice into a lever for product discovery, growth, and global scale.

New in the AI library - Create high density McKinsey-grade presentation decks

This is a template for generating high-fidelity strategy consulting slides that combines McKinsey-style rigor with premium editorial design. Product teams can use it to quickly create professional market analysis decks by simply inserting their target market (e.g., “US SaaS market,” “AI-powered healthcare”), removing the need to manually design complex layouts with data visualizations, matrices, and strategic frameworks. (Department of Product)

Strategy - Why Judgement is the most important product skill in 2026

Tech exec Gokul Rajaram helped build the core ads and product businesses at Google, Facebook, Square, and DoorDash, working directly with many of this generation’s best founders and CEOs. In this talk, he emphasizes judgment as the sole AI-resistant skill, drawing from experiences with leaders like Larry Page and Mark Zuckerberg.

Design - Stripe launched a new homepage

Stripe is known as an industry leader for homepage design. In this post, the company’s Head of Design, Katie Dill, shares Stripe’s brand new homepage which she describes as “chockfull of love”.

Tools you can use - Superagent from Airtable

Superagent is a standalone AI product from Airtable that acts like a multi‑agent “research and strategy team” for product decisions. For a PM, instead of manually stitching together market research, competitive analysis, financial modeling, and stakeholder briefs, you can ask Superagent questions like “Should we launch feature X in segment Y?” and it will plan the research, pull data from premium sources, run parallel specialist agents (market, finance, competitive, regulatory), and return an interactive, visual deliverable.

Anthropic’s CEO on AI and the “Adolescence of Technology”

This week, Anthropic’s CEO Dario Amodei published a 20,000 word essay on AI and his perspective of what the future of technology looks like.

Some of the important nuggets for product teams to consider include: a prediction that almost all code will be written by AI soon, guidelines on how to develop your own AI Constitution and a warning about the impact of AI’s “cognitive breadth” on the future of jobs.

More detailed analysis with real world examples here:

---

New product features and innovation this week

Google has unveiled a new technology called “Agentic Vision” in Gemini 3 Flash. You can read more about it here but it’s a new capability designed to improve Gemini’s image analysis. Instead of analyzing an image once and guessing at details it might miss, the model uses a “Think, Act, Observe” loop where it formulates a plan, writes and executes Python code to manipulate images (cropping, zooming, annotating, measuring), and then observes the results before providing an answer.

For product teams building AI-powered features where users upload artifacts this could improve the accuracy of those features. For example, e-commerce platforms could verify product listings by counting items and confirming the condition of an item with annotated proof, or insurance systems could assess damage claims by cropping and labeling each affected area.

Chrome’s Gemini Powers get deeper integrations with Workspace; ideas on what you can do with it

This week, Google also unveiled major updates to Chrome with the addition of Gemini in the sidebar, agentic auto browsing and the additions of Nano Banana . On the face of it, it looks pretty much like other AI browsers like Dia and Comet, but it comes with deep integrations into the Google Workspace VP of product Josh Woodward says right now he uses it to summarize multiple pages at once and then consolidate all of them in a single window.

Watch a demo of it in action here from Chrome’s engineering leader Addy Osmani where it collects information from a Google Doc, synthesizes it and sends the relevant information to recipients via Gmail.

Google’s Workspace integration makes this an AI browser that will be extremely difficult to beat. One commentator even suggested that OpenAI made a mistake by releasing Atlas because that opened the door for Google to avoid antitrust litigation by releasing the features as a defensive strategy.

How product teams could use Chrome’s new Agentic features

Here’s some of the ways product teams could use it:

• Usability test findings - a designer could have a Google Sheet open with user research findings and say “Email the design team highlighting the top 3 usability issues from this spreadsheet, grouped by severity.” Gemini extracts the data and drafts a structured update email.

• API version deprecation communication - a platform PM shutting down an old API could say: “We’re turning off v2 in 90 days. Check who’s still using it from the usage logs Sheet, match them to customer size in the CRM, then create three email templates: one for enterprise customers (white-glove, business-focused), one for mid-tier (webinar invite), and one for small accounts (automated, technical)

• Churn signal triangulation - a customer success PM could have NPS responses, usage data, and cancellation reasons in separate Google Sheets. If users citing ‘too complex’ jumped 40% this quarter, they could map these accounts against feature adoption in the usage Sheet – are they actually using advanced features and overwhelmed, or never adopting and bailing early?

Claude gets “interactive tools”; Asana’s CPO says it’s a fundamental shift in how people use software

Anthropic has launched “interactive tools” in Claude, meaning users can now use apps directly inside the chat interface without switching tabs. Some of the early tools available include Amplitude, Asana, Slack, Canva and Figma. Just before the Christmas break, ChatGPT did something similar with its own apps.

Unlike ChatGPT, though, Anthropic currently dominates the enterprise API market with ~40% market share. And this could mean that Claude is more likely to succeed in its vision to become the de facto workplace AI product; if companies are already using Anthropic’s APIs, asking employees to use Claude for their every day interactions with SaaS products seems more likely than OpenAI being able to convince them to use ChatGPT.

There are still some genuine UX questions around this interface. While completing some tasks inside Claude’s UI makes sense, for other, more intricate use cases it’s super frustrating and users are better off just opening the app they’re trying to use.

What product product leaders are saying about it

One of the early partners is Asana and their Chief Product Officer says that this represents a “fundamental shift” in how people use software. In his view, the chat interface becomes the execution layer where you get tidbits of info like status updates and insights - and the core Asana product is where you leave the chat interface to dig deeper.

And Intercom previously published a piece refuting the idea that vibe coded software will replace enterprise products, saying that while “Anthropic could have easily chosen to build an in-house AI agent to power its customer service… the team saw a greater opportunity to partner with a company with deep expertise in AI-first customer service so they could remain laser focused on their goal of building safe, reliable, and transformative AI for humanity.”

Microsoft Excel Agent Mode gets a wide rollout

Microsoft has released Agent Mode for Excel. It can create workbooks from scratch, perform scenario modeling, analyze large datasets, fix and generate formulas, and create data visualizations like pivot tables and charts - all through Copilot. Users can also switch between OpenAI and Anthropic models.

Here’s what Microsoft’s VP of Product for Copilot had to say about it and how you can use it:

---

📈 Product data and trends to stay informed

AI and vibe coding is accelerating the decline of software stock prices. Figma is down 80% since its IPO and new analysis in the WSJ shows Salesforce, Adobe, and ServiceNow have lost 30%+ of their value since early 2025.

The rise of Anthropic’s Claude Code and other vibe coding platforms means investors are spooked by the prospect that customers can build custom tools faster than they can negotiate enterprise contracts.

But not everyone agrees that this will happen. JP Morgan says it has no plans to replace enterprise tools like Jira and ex-Square CPO says it’s “foolish” to try to do so.

Pinterest is set to reduce its workforce by 15% according to its latest securities filing. Some of the cost savings will be used to support “transformation initiatives” that include but are not limited to prioritizing AI-powered products and capabilities. Some commentators have accused the company of so-called “AI washing” where a company exaggerates the impact of AI on its workforce to send a message of efficiency to investors and increase the stock price in return. If that is the case, Pinterest is certainly not the only public company guilty of it.

Gemini beats other models at long professional tasks in AI Agents. A new benchmark called APEX-Agents tests whether AI agents can carry out realistic, long, cross-application tasks in professional services (investment banking, consulting, and law).

The test involved:

• 480 tasks across 33 realistic work scenarios (researchers called these “worlds”)

• Tasks take professionals 1-2 hours on average to complete

• Each world contains ~166 files and 9 applications (spreadsheets, docs, email, etc.)

Gemini came top but the main takeaway is that even with the best models 75% still fail on their first attempt at more complex workplace scenarios. It’ll be fascinating to see how these benchmark scores change throughout this year.

Faced with losses in excess of $10 billion a year, OpenAI is experimenting with new product monetization models. Ads are coming with CPMs priced at a premium: according to The Information, the target price is $60 CPM versus less than $20 for Meta. As well as this, OpenAI will take a 4% cut on Shopify ecommerce purchases and, perhaps most interestingly, it is reported to be planning to develop a cut of scientific discoveries made using its new ChatGPT Science product, Prism, announced this week.

---

Paid subscribers get the full DoP Substack including: The Knowledge Series for sharpening your tech / AI skills, the AI Prompt and Skills library and DoP Deep dive reports for in-depth analysis to learn lessons from the world’s top tech companies.