2026 has been a weird year for me. I'm one of the most "pro DIY PC" people you'll ever meet. I built my first PC around 20 years ago, and I've never touched a prebuilt since then, except for the purpose of reviewing them. And yet, this year, I bought a prebuilt—not for me, but still.
GeekWire is profiling over the next few weeks some of the people and teams that are shaping the evolution of Microsoft in what we’re calling its “Microsoft 2.5” era.
Just Don’t Call It an ‘Agentic OS.’ Given Microsoft’s one-pointed AI focus these days, it’s not surprising that the Windows organization is on the agentic train.
But Executive Vice President of Windows + Devices Pavan Davuluri has learned the hard way not to call Windows an agentic OS. He did so back in November 2025, via a tweet and blog post, and the customer backlash was quick and biting.
But Davuluri has not done a complete U-turn because of the criticism. Instead, he has changed how he talks about where Windows is going — which is still in an agentic direction.
“The user of Windows going forward will continue to be users … but it’s also going to add these agentic workloads,” the nearly 26-year Microsoft veteran Davuluri told GeekWire in a recent interview.
During his time at Microsoft, he’s held a variety of roles, from intern to General Manager of Surface, to Corporate Vice President of Windows Silicon & Systems Integration. He was appointed Executive Vice President of Windows + Devices in March 2026, reporting directly to CEO Satya Nadella.
Windows needs to evolve to support agentic workloads through new platform capabilities that the team is building under the covers, Davuluri said. These low-level capabilities, or “primitives,” affect how Windows handles security, identity, governance, observability, and performance when it comes to building and running agents natively.
These coming changes likely will affect the Windows file system, security model, PowerShell, and other foundational components.
Microsoft already is working on Windows identity and manageability to make them better able to service agents. Windows can assign agents a local ID, or a cloud-provisioned identity backed by Entra.
And it also has an early preview of technology known as Microsoft Execution Containers, meant to help secure agents by running untrusted code in sandboxes or virtual machines. It’s these system-level areas where the team is focusing first in preparation for a human+agent future, Davuluri said, rather than the UX/UI level.
Going Back to Basics. Windows has had a lot of very different leaders over the years, with very different management styles and priorities.
For his part, Davuluri said he plans to run the Windows and Surface teams with four principles in mind: Maintaining customer obsession; treating Windows as a complete end-to-end system (“full stack”); focusing on complete user experiences and workflows rather than individual features; and building Windows openly and transparently, with clearer communication about plans and priorities.
On the heels of his promotion to EVP, Davuluri committed publicly to the much-needed goals of improving Windows quality and reliability. In a blog post, he outlined some of the requested changes that his team would be making to Windows, ranging from fixing the way the Insider test program works, to more granular improvements like allowing users to reposition the Windows task bar.
And since then, the team largely has been delivering to the surprise and delight of many long-time Windows users.
Davuluri has also been working to shift the conversation from which new features are coming to a specific build to what are the outcomes Microsoft wants to enable for specific groups of Windows users.
“There is no one single sort of ring for a billion-plus users on the platform,” Davuluri said. Windows users encompass people who use the product in a variety of different ways, so “we need to get clarity in our minds on the things that we do that lift all boats that raise the entire platform — and things that we have to go do that are specific and unique to each of our sets of users based on how they primarily or typically use the device.”
Full-Stack Thinking. Is there still a role for Microsoft as a PC maker in the coming agentic future? Not surprisingly, given his heavily hardware-focused background, Davuluri insisted there is.
When Microsoft debuted its first Surface devices in 2012, officials said the company needed to build its own hardware to create reference designs and innovative form-factor examples for other Windows PC makers.
These days, most Surfaces that ship arguably are not better, spec- or design-wise, than other PCs. But Microsoft still needs to keep a hand in hardware design to understand the full stack, Davuluri claimed.
Surface plays a key role in how Microsoft develops platform abstractions, incubates support for technologies like pen, facial-recognition, and neural-processing units that later spread across Windows, and optimizes for silicon-to-cloud, he said.
While the company’s attempt to create a distinct category of “Copilot+” AI PCs fizzled, Microsoft continues to try to find AI-centric reasons to convince customers to choose Windows devices. Davuluri and others have referred to the idea of “unmetered intelligence” to attempt to make the case for running AI models locally on PCs.
This fall, Microsoft (and other Windows PC makers) plan to roll out new PCs built on the Nvidia RTX Spark platform. The coming Surface Laptop Ultra, which will be optimized for RTX Spark, is aimed at creators, developers and AI builders, all of whom — Microsoft is hoping — will be fueling the growth of its next target user category: Agents.
The commander and pilot of NASA's Artemis II mission have gained a new distinction five months after splashing down from the Moon: They're not fully retiring from the space agency.
Reid Wiseman and Victor Glover are the first astronauts to join NASA's emeritus program, enabling them to continue to support the work being done at Johnson Space Center in Houston. They will remain available to train and mentor the current workforce while still being able to pursue employment and opportunities outside of NASA.
"I have asked NASA ... to not use the term 'retire' as much, but technically emeritus is a retirement program," Glover told collectSPACE.com on Wednesday. "It's something we have typically done for scientists when they have groundbreaking research, and they want to go back to academia to teach or to research and publish."
© NASA/Kim Shiflett
Meta took days to remove ads containing AI-generated child sexual abuse material (CSAM) on Facebook and Instagram. Some ads featured photos of real kids, including a press photo of a young member of a European royal family and images swiped from a popular Instagram profile of a preteen girl deemed an influencer.
In an investigation published Tuesday, the Tech Transparency Project (TTP) reported that Meta failed to detect 332 ads containing CSAM this year. The “vast majority” of ads promoted AI apps made in China, while many ads promoted so-called “nudify” apps that make it easy for bad actors to use AI and digitally alter images of children.
TTP matched “multiple CSAM ads to photos of real children that appeared online.” These ads seem to violate federal child pornography laws, since the Justice Department has clarified that AI CSAM is just as harmful as CSAM. The young royal’s image was “animated into a video of her performing a graphic sex act,” TTP found. Other ads animated a photo of a 14-year-old Instagram influencer “showing off her new sports club uniform” into “a video of her performing oral sex.” A third “preteen” victim “posing in a pink athletic outfit with pigtails” in a series of stock photos was morphed into a video where she looks frightened as she’s molested by an adult male, TTP reported.
© Alex Wong / Staff | Getty Images News
Instagram will limit recommendations for AI-generated influencer profiles that fail to disclose their synthetic identities under a new labeling policy.
The post Instagram Cracks Down on AI Influencers That Pretend to Be Real appeared first on TechRepublic.
In the business world, a moat is a quality of a business that makes it difficult for competitors to take that company’s profits. With how hard it is to train models for large language models (LLMs) and generative AI, it might seem like Anthropic, Open AI, and other LLM companies would have huge moats given the amount of compute it takes to build models. But open source models are quickly draining that moat, and now the only thing standing in the way of a customer using one of these models on their own hardware instead one from the larger companies is physical computing resources. [TerminalBytes] demonstrates a few of these models on personally owned computers to show the current state of the art.
[TerminalBytes] started off running the 27B version of the Qwen3.8 on a Mac Studio with 256 GB of unified RAM, which is plenty for this task. But it’s also enough to benchmark a few different models. Qwen3.6 is compared to 3.8, and then the different quants of each model are also compared. Quants are compressed versions of models that need fewer bits to store weights, meaning that the same models can run in less memory with smaller losses in fidelity. Many of these quants run on machines with 32 GB of RAM or less, encompassing many average gaming PCs. There’s even a 1-bit quant that [TerminalBytes] tested which can easily run on a machine with 16 GB, although with mixed results.
Keep in mind that this is just the current state of affairs with open LLMs. Future versions of these models are likely to optimize the number of tokens produced per unit time, or otherwise increase quality of responses while requiring less computer resources. We don’t really think that the ease of running local models will be the sole reason that the AI bubble pops, though. The fact that not every computer user is running Linux is proof enough of that.
We all know that the price of RAM and storage has shot up due to demand from the AI industry and other factors. In most cases that means we grumble about the price, but if we really need the part we can fork out for it. [CNX Software] are reporting that rather than continue to push up their prices, Pine64 are responding to the crisis by halting production of their Linux boards for the time being.
We’ve seen online comment over the now-exorbitant cost of other boards such as a fully-loaded Raspberry Pi, and this follows in that vein. If we had to guess we’d speculate that the high process have resulted in too little in the way of sales, which considering the knock-on impact on our community if other vendors follow suit, could be concerning.
If there’s one bright spot in the current situation, it’s that for many applications where a single-board computer might be used, a microcontroller might now be a better choice for the job than something running Linux. We’re in a very different situation from that we were in when cheap Linux boards appeared, the current generation of high-power microcontrollers have significantly closed the gap between the two. Given that microcontrollers have onboard memory and storage, their immunity from the price hikes makes them even more attractive. As to Pine64, we hope that sales of their other products make up for it.
Header image: BasilicumTree, CC0.
Meta agrees to pay up to $18 billion and introduce new Facebook and Instagram limits for teens under a US state settlement over addiction claims.
The post Meta Settles Teen Social Media Addiction Case for Up to $18B appeared first on TechRepublic.
Meta agreed to impose daily limits on children's social media use and pay nearly $18 billion in settlements with nearly every US state today, cutting short a trial in which Meta said several of the states were demanding over $1.4 trillion. The settlement requires court approval.
Meta is facing claims that it designed its products to foster compulsive use by children and failed to warn users of addiction and mental health risks. Meta, which already uses ID checks and face analysis to verify user ages, said it agreed to impose on people under 18 a "default two-hour daily time limit that teens can only turn off with a parent’s permission," a default block between midnight and 6 am, and a school mode in which notifications are muted by default from 8 am to 3 pm.
The two-hour daily "limit is cumulative across Facebook and Instagram, and time spent scrolling on both apps counts toward the total, including if we detect that someone has multiple accounts," Meta said. Teens will "receive prompts after every 15 minutes of continuous screen time on Facebook or Instagram," and "prompts when their total daily usage hits 60 minutes and 90 minutes."
© Getty Images | NurPhoto
Washington state will receive up to $339 million as part of a historic $17.1 billion multistate settlement with Meta, resolving allegations that the tech giant intentionally designed Facebook and Instagram with addictive features that harmed youth mental health.
Attorney General Nick Brown said Wednesday that the landmark agreement delivers on core youth-safety product changes — including hard caps on daily time limits, late-night scrolling blocks, and disabled push notifications during school hours — that state lawmakers failed to pass through legislation over the past two years.
“Let me say to the young people of Washington state: This agreement shows that your health and safety is more important than Meta’s profits,” Brown said in a news release.
Under the deal, Meta must overhaul Instagram and Facebook for young users by enforcing a two-hour combined daily time limit, turning off push notifications during school hours (8 a.m. to 3 p.m.), and blocking access late at night between midnight and 6 a.m. Teen users will also get the option to switch off algorithmic feeds in favor of a chronological timeline.
The agreement resolves claims brought by a coalition of 47 states, Washington, D.C., and three territories. While Meta acknowledged the settlement could cost up to $18 billion total over 10 years, it marked a rare legal resolution for a major platform facing nationwide youth safety litigation.
Outside the landmark Big Tobacco agreements of the late 1990s, the $17.1 billion deal represents the largest state consumer protection settlement in U.S. history. State officials and tech policy experts are framing the enforcement action as a similar watershed moment for regulating algorithmic harms and digital product design.
Washington state will receive a guaranteed baseline of $237 million from the core youth-safety agreement, with its payout potentially scaling up to nearly $339 million over the next decade if other major platforms like TikTok and Snapchat adopt comparable terms.
The Attorney General’s Office plans to use the funds to cover legal costs, bolster ongoing consumer protection enforcement, and directly fund state programs tackling the youth mental health crisis driven by social media use.
The settlement also mandates an independent third-party auditor to evaluate and report Meta’s technical compliance directly to state regulators annually over the next five years. Beyond usage limits, Meta must restrict social comparison features such as targeted beauty filters, hide public “like” counts for younger users, and implement stricter age verification to prevent children under 13 from creating accounts.
In addition to the core youth-safety agreement, Washington will receive a separate $10.2 million payment resolving long-standing state claims against Meta for sharing nonpublic user data with third parties like Cambridge Analytica during the 2016 election cycle.
Oregon Attorney General Dan Rayfield announced that his state’s share will total more than $125 million over 10 years.
In a public statement, Meta praised the agreement as setting a new benchmark for youth safety, while emphasizing that the restrictions should apply across the entire industry.
“While this is an important step, these protections will only be truly effective if our peers — TikTok and YouTube — put the same measures in place,” a Meta spokesperson said.
The agreement remains subject to final judicial approval in federal district court.
The fourth annual Seattle Tech Week got off to a big start Monday, with panels and parties bringing together thousands of people from across the region and out of state. Organizers said the week features more than 250 events and drew more than 29,000 event registrations.
We went to Madrona’s kickoff event at Picklewood Paddle Club with one question for the founders, investors, and operators we met: What are you building? Here’s what we heard and learned.
What he’s building: An AI system that runs a professional cricket franchise — flights, hotels, ground transportation, and daily schedules for players and staff, all handled over WhatsApp.
Nemani is chief product officer of the Seattle Orcas, the Major League Cricket team now in its fourth season. For the first three, he ran team operations the old-fashioned way: “I ran the entire operations using spreadsheets and people and processes,” he said. That meant tracking a constant stream of inbound flights, hotel blocks and car bookings across a season.
This year, he used Claude Code to build the backend for an AI agent that took over roughly 80% of the operation: booking flights, hotels and cars, dealing directly with hotels and transportation vendors, and telling players and staff when their flight lands, which hotel they’re in, and who’s picking them up. It also handles daily schedules, down to massage appointments.
To accommodate players and staff who were reluctant to adopt new tech tools, he built it to run on WhatsApp, the messaging app they already used every day.
What she’s building: A B2B tool that lets thrift, vintage, and consignment resellers photograph an item and get back the identification, pricing, and listing details they now assemble by hand.
Vu is founder and CEO of StyleOrigin. Getting a single secondhand garment listed for sale is still manual work that takes 30 to 45 minutes an item, she said. With StyleOrigin, a reseller takes one image and an AI analysis returns what they need to list and price it. The company also gives sellers data to guide inventory decisions.
She found the problem herself. Vu ran environmental, social and governance work at Remitly until she stepped down in 2023, then took a year off and started selling vintage clothing. She assumed she was slow because she was new to it. “But turns out everybody does it the same, and so there wasn’t really any good solution out there.”
She taught herself to code and built the first version of the product. StyleOrigin has a working MVP but no revenue yet. More than 70 stores around the country are on a waitlist, and Vu is about to bring her first engineer aboard.
What he’s building: Tools for collecting, storing, and analyzing the data AI systems produce — the record of what agents actually did, not just the code they shipped.
Daniel is founder of Hyperparam, an early-stage Seattle startup, and previously co-founded Algorithmia, the Seattle machine learning company acquired by DataRobot in 2021.
Companies are spending heavily on AI without much sense of what they’re getting, he said. “AI is producing this wall of tokens. Companies are paying huge amounts of money to generate all these tokens, but they have really no visibility into what are these agents doing.”
Every token leaves a trail, and Daniel said most companies ignore it. Mining it would show them where AI is working and where it’s wasting money.
“Where are models being stupid? Where are they going down rabbit holes?” Older analytics tools can’t help, he said, because they were built for numbers and clicks: “People haven’t really been thinking about what do you do when the majority of the data being produced in the world is text.”
What she’s building: An urban mine — recovering precious metals from jewelry and returning them to the supply chain for clean technology.
Escarez is founder of Redyoos, which GeekWire featured in Startup Radar last year. The jewelry industry accounts for 40% to 50% of the global supply of precious metals, she said — the same materials found in “anything that has an on and off button,” from cell phones to wiring.
Demand for those metals is climbing with AI and clean energy, and Escarez said projections point to a supply shortfall of 700% over the next couple of decades. “We mathematically cannot solve this deficit,” she said, which is why she sees jewelry as a viable source.
Redyoos collects jewelry, refines what contains precious metals, and sells the recovered material to clean-tech manufacturers.
Escarez, a former chief operating officer at Boma Silver Jewelry and brand manager at Starbucks, has bootstrapped the company, which has been live a little over a year and is generating revenue. She is now raising a pre-seed round.
What he’s building: An engineering team inside a venture capital firm, automating the work of investing.
Liu is a partner at Unlock Venture Partners, which he helped launch in 2018 to back early-stage startups in Seattle and Los Angeles, and which raised a $60 million second fund in 2022. A longtime Seattle entrepreneur and angel investor with stakes in close to 100 companies, he was previously CEO of BuddyTV, acquired by Vizio, and of NetConversions, acquired by aQuantive.
“We actually have an engineering team that’s trying to automate a lot of what we do in VC,” Liu said, “and trying to make sure we can scale our business just like our own portfolio companies.”
The work covers deal memos and diligence on prospective investments, along with the mechanics of dealing with the firm’s own investors and collecting updates from portfolio companies.
The point, he said, is better decisions: “How do we get smarter as VCs?”
What she’s building: Private AI — letting people own their own data and context, use any large language model, and not be tracked or trained on.
Jesse is co-founder and CEO of ACME Brains, whose first product, nexie, is in beta. GeekWire wrote about the origins of the company last year: after her husband passed away, she turned to ChatGPT and found real comfort in it, then ran into its limits — it couldn’t carry the context of their conversations, and she had concerns about the privacy of what she was telling it.
nexie keeps a user’s notes, journals, and conversations in what the company calls a personal context engine, and carries that context across AI services instead of leaving it scattered in separate chat histories.
Trading privacy for free services goes back to the early internet, she said, but AI tilts the exchange further. A chatbot draws information out of a person in conversation, then combines it with everything already known about them. “AIs can talk you into your data,” she said.
An electrical engineer with more than two dozen patents who spent decades in wireless at McCaw Cellular and AT&T Wireless, Jesse said most people don’t grasp how AI actually behaves, which leaves them exposed — seniors especially. “You need people that understand it to help protect people that don’t.” Her co-founders are Alan Caplan, Amazon’s original general counsel, and patent attorney and engineer Bob Bergstrom.
What she’s building: Voice AI that lets bar and restaurant staff count inventory out loud instead of writing it down by hand.
Rapp is founder and CEO of Köniva. A typical hotel resort bar spends 12 hours and four people on an inventory count, she said; with Köniva it’s two people and 3-and-a-half hours, and more accurate. Staff download an app and wear a lapel mic — you want both hands free on a ladder — and count out loud the way they always have.
She came to the problem after a career in big tech and ad tech. Not wanting to build for an industry she’d never worked in, she took a part-time job at Canlis after training as a sommelier.
When she was injured, the wine director let her help with inventory reconciliation and handed her a clipboard of handwritten numbers plus a login to the restaurant’s inventory software. She asked why they were still using paper and pencil when a whole engineering team had built software for the job. The wine director’s answer: it was faster.
Köniva has 10 customers. At several high-end hotels and restaurants, Rapp said, staff put the app on their personal credit cards to start using it, then helped her pitch their own procurement departments — an unusual path in an industry she said has been badly burned by technology.
“It is insane how bad tech has been to them,” she said.
What he’s building: A growth equity firm investing at the intersection of food brands and food tech.
Arias is founder and managing partner of Altelan Capital, a Seattle firm he started last year. It underwrites companies around the Series A stage, generally, providing growth capital and strategic support.
He came up in the industry itself, leading finance at restaurants and breweries and most recently running corporate development and financial planning for Seattle Hospitality Group. That operator lens, he said, is what he brings to investments and to coaching founders on growth. He has been an investor since 2015.
Arias calls Altelan an AI-native investment fund, using AI tools to get up to speed on an industry and test assumptions about a business’s ability to scale and where the risks are. He’s equally interested in where the technology doesn’t belong and simplicity is the better option: “AI is great, but it may not be the right tool for the job.”
The bigger shift he’s watching is food and digitization. The industry has traditionally worked off “the proverbial clipboard and a notepad,” he said, and the pandemic accelerated the move to technology across the supply chain. “There are many applications of tech in food,” he said, “and that’s what keeps us up and gets us excited every day.”
Login