What You Should Know
AI’s Busy Week Points to a Deeper Role in Everyday Work
AI news tends to clump together, but the past week was on another level. Anthropic announced a $15 billion strategic partnership with Microsoft and Nvidia. OpenAI expanded its enterprise footprint through a new $100 million deal with Intuit. Nvidia will report earnings this afternoon, with investors questioning whether the AI build-out is peaking.
Aside from the Monopoly money, there were also big product announcements: OpenAI updated its flagship model with GPT-5.1, Google deployed Gemini 3 — including in its search engine’s AI Mode, which has 2 billion monthly users — and Microsoft announced an aggressive push into AI agents at its Ignite event.
These three releases point in the same direction. AI is evolving from generating text and code to becoming a more helpful and capable assistant. Let’s break each one down.
GPT-5.1: Better control, finally
OpenAI didn’t position GPT-5.1 as a monumental upgrade, but early tests show it quietly fixes a big problem with GPT-5: instruction-following. In the model release’s accompanying blog post, OpenAI shows how GPT-5 wanders off a simple rule to “always respond with six words,” while GPT-5.1 adheres and doesn’t lose much color despite the brevity.
TechRadar pushed it further with a Lion King test: four sentences, clear enough for a 7-year-old, no sentence starting with “Simba” or “The.” GPT-5 broke the “the” rule. GPT-5.1 followed every constraint and squeezed in character names and plot detail without getting more complex. For anyone writing under tight word counts, picky clients, or both, that kind of discipline is what makes an AI model most valuable.
The other big shift is tone control. GPT-5.1 has richer personality presets inside ChatGPT — Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical — and they actually stick. GPT-5 had similar presets, but they often faded halfway through a thread. For anyone juggling client voice guidelines or strict editorial tone, consistency is the upgrade that matters.
Gemini 3: Leaning into multimodal
CEO Sundar Pichai called it the company’s “most intelligent model, that combines all of Gemini’s capabilities together so you can bring any idea to life.” Google’s differentiator is its multimodal capabilities, and Gemini 3 “significantly outperforms [Gemini] 2.5 Pro on every major AI benchmark.”
Hands-on tests back up the positioning. In Tom’s Guide’s evaluation, Gemini 3 outperformed GPT-5.1 in coding, format compliance, and image interpretation. That matters for communicators because so much of the job isn’t pure writing, but working across formats.
What’s more, Gemini 3 includes Generative UI, which creates “immersive visual experiences and interfaces,” like web apps, games, and graphics, that are customized in response to any prompt. Where most AI tools would only be able to parse data and provide written storylines, Gemini 3 could tell the story graphically.
Microsoft Ignite: Agents everywhere
Microsoft went all-in on AI agents — autonomous or semi-autonomous tools that can take actions inside Microsoft 365, Teams, the Office suite, and enterprise apps.
Microsoft introduced Agent 365 as a centralized system for deploying, managing, auditing, and governing AI agents inside organizations. While communicators might not be the core audience for this release (it’s really geared toward CISOs, CIOs, or IT teams responsible for overseeing AI implementation), it’s a signal that organizations are increasingly turning to AI agents to automate entire processes.
Even if you don’t work closely with an AI agent yet, now is the time to start thinking about how you’d want one to work alongside you. Which tasks take up too much time that you’d want an agent to do for you? Which tasks would you want to keep for yourself?
The three big releases hint at a workplace where the dull parts of communication disappear into software and the human work shifts upstream. Everyone in comms and PR may feel that change sooner than they think.
Elsewhere …
- LISTEN: Small Businesses Deserve Capital — and AI Can Deliver It
- Google Updates Its Weather Forecasts with a New AI Model
- Bloomberg Uses ‘Synthetic Characters’ to Test Out Products and Marketing
- AI Data Centers Are Banging on Crypto’s Door
- Target Announces Partnership With OpenAI as It Aims to Reverse Sales Slump
- Fortune 500 Scrambles After Anthropic’s Warning of Automated Cyberattacks
Tips and Tricks
The apps are here
What’s happening: Early last month, OpenAI held its DevDay and announced that apps would be coming to ChatGPT. Well, now they’re here.
How they work: If you go to settings, you’ll see “Connectors” is now “Apps & Connectors,” with nine initial apps: Booking.com, Canva, Coursera, Expedia, Figma, Peloton, Spotify, Tripadvisor, and Zillow (Tripadvisor and Peloton are new additions to the first seven that were previously announced).
When you click on one and hit the “Connect” button, you’ll see some fine print that includes a section titled “Data shared with this app.” Be sure to read that and make sure it aligns with your organization’s AI policy. Each app includes this line from OpenAI: “Our policies require that apps only access relevant content to respond to your requests.”
Which apps work best: Of these initial nine, the two that stand out as benefiting communicators are Canva and Spotify. The latter is more about finding the tunes that help you focus. You can converse with Spotify through ChatGPT to build playlists or describe what you’re looking for, which is more helpful than just using the search function in Spotify, where you’d need to know the name of the song or artist you’re looking for.
Canva can help a bit more concretely through creative work. You can connect your Canva account with ChatGPT to work on new designs or update existing ones. However, if you’re looking to build an image, you will use Canva’s image generator, not ChatGPT’s. If you prefer ChatGPT’s image generation skills, you’ll have to ask for that specifically and upload the image to Canva manually.
Quote of the Week
“We use AI in three ways. Number one is to help our customers, and that’s the thing we find most exciting and interesting.
“We’ve added AI summaries on to our articles. We’ve used AI to improve our search engine, and that has led to 15% click through improvement.
“We also use AI to help our business. So we’ve used AI and LLM technology to build out our contextual targeting [advertising] offering, which has been really important to our video revenue growth.
“Using this technology to understand contextually what’s happening in a video, and provide that alignment for our marketing partners.
“And then [the] third way we use AI is internal workflows.”
— Julia Beizer, Chief Operating Officer at Bloomberg Media, speaking at Press Gazette’s Media Strategy Network event
How Was This Newsletter?
