📊 Claude Goes Live Across Microsoft 365

Anthropic is now competing with Microsoft inside Microsoft's own apps. Claude went generally available across Excel, PowerPoint, and Word today, with Outlook in public beta, even as Anthropic's models continue powering parts of Copilot in the background. The model layer and the workflow layer are starting to drift apart. Let's dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: Claude is now generally available across Excel, PowerPoint, and Word, with Outlook in public beta.

Threads now move between Outlook, Word, Excel, and PowerPoint without losing context, so an email can lead into a memo, a model, and a deck in one continuous workflow. A change to an Excel assumption flows into the chart in the deck and the figure in the memo. Edits surface as tracked changes in Word, highlighted cells in Excel, and unsent drafts in Outlook, keeping a human in the loop before anything ships. One AppSource listing covers Excel, PowerPoint, and Word, Outlook gets its own.

Anthropic models already sit inside Microsoft Copilot as a subprocessor, and now Anthropic is shipping a parallel direct-to-user product in the same apps. It's a strange dual position to hold. The model layer is commoditizing across surfaces, and the real fight is shifting toward the workflow that knowledge workers live inside.

Go on tangents. Change your mind mid-sentence. Say "um" twelve times. Wispr Flow doesn't care — it takes everything you say, strips the filler, and gives you clean, structured text ready to paste into any AI tool.

The result: prompts with the full context your AI tools need to give you useful answers. Not the abbreviated version you'd type because typing is slow.

Works inside ChatGPT, Claude, Cursor, and every app on your screen. Millions of users worldwide, including teams at OpenAI, Vercel, and Clay.

Start flowing free

Evolving AI: Mozilla used Anthropic's Mythos to patch 271 Firefox bugs, redrawing how AI fits into software defense.

Mozilla just shipped Firefox 150 with 271 patches sourced through Anthropic's Mythos model, and several of those flaws had quietly survived years of human review and traditional fuzzing, some buried in the code for over a decade. The pace tells the story too, bug volume jumped from 31 fixes in April 2025 to 423 in April 2026. Even with that surge, Mozilla kept humans in the loop for the actual repairs, with one engineer writing each patch and another reviewing it. Mythos handles discovery, people still handle the fix.

Cybersecurity has always tilted toward whoever finds the bug first, and for decades that was usually the attacker. Mozilla's haul nudges the geometry a bit, since AI scanning legacy code at scale finally starts chipping away at the backlog defenders inherited from the pre-AI era. The harder question is what happens once the same capability turns into everyday infrastructure on both sides of the fence.

The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.

Learn more

Evolving AI: Google DeepMind is taking a stake in the studio behind EVE Online and turning the 23-year-old space MMO into an AI research lab.

Google DeepMind is moving into one of the strangest AI research environments ever proposed. EVE Online is a 23-year-old space MMO with a player-run economy, corporations that last for years, and wars fought over real resources. The studio behind it even employs a former Central Bank of Iceland economist to manage in-game inflation. DeepMind will run an offline copy on a local server to test how AI models cope inside that kind of world. The partnership lands alongside a separate but linked move, with the studio buying itself back from Korean owner Pearl Abyss for $120 million and rebranding from CCP Games to Fenris Creations.

DeepMind has been climbing a difficulty curve for a decade. Atari, then Go, then StarCraft, and last November SIMA 2 figuring out 3D games it had never seen. EVE jumps the curve again because nothing is scripted, the economy is real, and decisions play out over years. Sociologists and economists have treated it as a research site for ages. AI is just the latest discipline showing up to study what humans do when no one writes the rules.

QUICK HITS

🧠 Use AI without outsourcing your thinking to it.

🌙 Moonshot AI hits $20B as open-source demand soars.

🧠 Fruit fly brains could power the next AI breakthrough.

🎙️ Spotify lets AI agents create personal podcasts on demand.

 *partner link