Welcome back! Companies are now recruiting internal “AI champions” to convince skeptical coworkers to actually use the AI tools they’re paying for. A new BCG report found that 74% of front-line white-collar employees now use AI daily or several times a week, up from 51% last year.
Turns out, the hard part of enterprise AI may not be buying the software. It’s getting humans to change how they work.
A Big Announcement (And a Milestone I Can Barely Believe)
A few days ago, my YouTube channel turned 17 years old. And I'm about 24,000 subscribers away from 1 million. Wild.
I could not have imagined in my wildest dreams what this channel would eventually grow into. I get to spend my days trying cutting-edge tech, breaking down AI news, and nerding out with a community of people who care about the same stuff I do—and somehow this is my job. Beyond grateful.
To celebrate, I've got an announcement I've been sitting on for a while, and it's a big one: FutureTools.io is transforming.
For years, FutureTools has been a wide AI news and tools repository—basically a "here's everything" firehose. That was useful for a moment, but the AI space has moved so fast that "everything" has become impossible to keep up with. So we're pivoting the site into a hub for curated AI knowledge, industry insights, exclusive commentary, and handpicked tool reviews.
The big news and tools lists are still there. But the focus is shifting from "here's everything" to "here's what actually matters." Just the stuff you need to be the smartest person in the room on AI, in one place.
This is the culmination of an ethos I've been carrying for a while: AI can help with a lot of things, but it can't replace taste. Yes, AI helped me write summaries and even helped me build most of the new site. But everything you see on there was curated by me, based on what I think is genuinely noteworthy.
Go check it out and let me know what you think. And if you want more curated insights delivered to you every week, you can subscribe to the newsletter right on the new site.
Can't wait to see what the next year (and the next million) has in store.
Thanks for being part of this ride.
— Matt
Trump Administration Clears OpenAI’s GPT-5.6 Launch
Via TechCrunch
GPT-5.6 is officially live. The flagship Sol model and lower-tier Terra and Luna models went public yesterday, after the Trump administration gave OpenAI the green light.
How we got here: GPT-5.6 had been under government review because of concerns about how powerful frontier models should be released, who gets access, and whether national security officials should have a say before the most advanced systems go public.
What changed: This week, the administration cleared OpenAI for a broad launch, suggesting the company and government officials found a path forward. Details are still limited, but the decision marks another case where frontier AI releases are being negotiated in real time rather than governed by a clear, settled framework.
What’s new: OpenAI says GPT-5.6 is more capable than previous GPT models. The rollout gives OpenAI another chance to widen the gap against rival labs while bringing its newest model family to the public.
Why it matters: This is another example of the murky line forming between AI labs and the government. Frontier models are increasingly treated like strategic technologies, but there still isn’t a clear rulebook for when Washington should step in, when companies should be allowed to ship, and who gets to decide what counts as too powerful.
Meta’s New Image Generator Is Already Making People Uneasy
Meta launched Muse Image, a new AI image generator from Meta Superintelligence Labs, and users are already pushing back over how it can use people’s photos.
What it does: Muse Image is available for free through the Meta AI app, Instagram Stories, and WhatsApp. Like other AI image generators, it can create images from prompts, edit existing photos, generate custom ads, test interior design ideas, and apply AI effects to Instagram Stories.
Why people are mad: One feature lets users manipulate another Instagram user’s images with AI as long as that person’s profile is public. Users can tag someone, pull in their photo, and generate a new AI image from it. Meta says people have controls to disable this, but users are not automatically notified when AI content is created using their Instagram content.
Privacy baggage: The backlash is hitting harder because it’s Meta. The company has a long history of privacy controversies, from the Cambridge Analytica scandal to shutting down Facebook’s facial recognition system after lawsuits and regulatory pressure. An opt-out AI photo feature gives users yet another reason to distrust the company.
The bigger picture: People are not rejecting AI image tools outright, but they are drawing a line around consent. Using AI to make a goofy image, mock up a couch in your garage, or edit your own photo feels very different from letting strangers on the internet remix your face.
Hear How Engineers Are Making AI Work for You
Working Smarter is back for season three. Hear how Dropbox engineers have been building context-aware intelligence that connects to all the tools your team uses for work—so you get AI that works wherever you do.
With episodes on context engineering, multimodal search, agentic AI, security, and more, Working Smarter goes behind the scenes to show you what it takes to build AI that actually understands you—and how it can help you work smarter, too.
Ready to hear the inside scoop? Listen to Working Smarter wherever you get your podcasts, or visit workingsmarter.ai
Turn Your Favorite Experts Into an AI Advisory Board
Via Adviserry
Adviserry is an AI-powered advisor that turns the newsletters, YouTube channels, and documents you follow into a searchable knowledge base and virtual advisory board. It continuously scans new content, synthesizes recommendations across sources, and gives you copy-paste action drafts tailored to your projects.
How you can use it
Build a searchable archive from creators and experts you trust
Get answers grounded in your chosen sources
Receive cross-source recommendations for your projects
Turn unread newsletters and videos into action drafts
Pricing: Paid
Site Observability for AI Agents
Via askbowtie
askbowtie is a lightweight site observability MCP that connects analytics, errors, search, ads, and uptime into one per-site data layer your AI agent can query. Agents like Claude Code, Codex, Cursor, or any MCP-compatible tool can ask plain-language questions and get cited answers about conversions, errors, ad performance, and uptime.
How you can use it
Let AI agents troubleshoot site issues with real data
Connect Google sources into one observability layer
Get digests about conversions, errors, ads, and outages
Replace dashboard-hopping with conversational reporting
Pricing: Free and paid plans available
Jobs, announcements, and big ideas
Google ships AlphaEvolve, its AI code-optimization agent, to general availability for all cloud users.
SpaceXAI debuts its Grok 4.5 model for coding and agentic tasks.
Microsoft uses audio AI models to bring classical Ode poetry to life in a new project.
OpenAI’s second-in-command Fidji Simo steps down.
Anthropic resets Claude's 5-hour and weekly rate limits for every user.
OpenAI introduces ChatGPT Work Agent to manage apps, files, and long-running projects.
Mistral launches Robostral Navigate, an 8B model that navigates using just a single camera.
Another huge week in the world of AI. I break down the highlights ⤵️
That’s a wrap! See you next week for more.
