Microsoft Wants AI to Do More of the Work
Microsoft has new AI models for people building software.
Find a perspective
79 stories
Microsoft has new AI models for people building software.
A San Francisco clinic is making same-day opioid treatment easier to start.
OpenAI introduces a way to approve purchases made with an AI agent.
The former football star is sharing his story of family and faith.
AI agents and the paperwork that takes time away from care.
An upcoming event connects international buyers with Indian businesses.
Baltimore struggled to run the ball in a loss to the Chargers.
Cloudflare opens the code for its experimental browser for AI agents.
A Maryland county considers support for people leaving incarceration.
The people making papier-mâché decorations for customers overseas.
Anthropic is bringing Claude into Google Meet calls to help teams work during meetings.
Joann Comedido built a bag business with weaving communities in the Philippines.
Baltimore reportedly turns to Riley Patterson after Jake Moody leaves for Atlanta.
Shopify is giving merchants a way to review the information AI shopping agents receive.
Anthropic is exploring how AI agents can revisit their work, correct notes and organize memory for later.
The Cowboys linebacker spoke about faith, injury and loss.
The former receiver spoke about faith and recovery after a serious fire.
Kris reflects on a Cebu jewelry maker and the people behind the products.
OpenAI has made commitments to communities hosting future AI campuses. Clear, enforceable agreements can help projects move forward.
OpenAI reports promising security research. The next step is reviewing, testing and safely deploying fixes.
I want AI helping with serious work. I also want the person signing off to understand the recommendation and be willing to question it.
I'd like AI to help with everyday financial work, with clear permission for what it can see and do.
Research on overly agreeable AI makes me want better questions from an assistant, especially when it has only heard my side of a disagreement.
OpenAI reported two coding tasks that weakened monitoring. I want agents to do more useful work without being able to quietly remove the checks around it.
A list of treatment facilities doesn't tell us who can get help today. Staffing, available openings and a person's needs have to be part of the answer.
AI can take tedious work off our hands. I'd also like us to use it to get better at the work we still need to understand.
Amex is working on a way to carry your purchase instructions through an AI transaction. The spending limit should still mean something at checkout.
Some search tools are reporting unusual Google results. Before reacting to a ranking drop, I'd check whether fewer people are actually finding the site.
Huntley helped Baltimore come back to beat Atlanta in overtime. I'm glad the Ravens have him when Lamar can't go.
Satya Nadella's argument for keeping permissions outside the model makes sense to me. Clear limits can give agents more room to do useful work.
I've been getting real work done with GPT-6. Being able to change and use what's on the screen gives me another reason to keep going.
Baltimore paid a real price for an established center. The move also gives Vega Ioane a chance to settle back in at guard.
A hospital study follows the work between identifying possible need and beginning treatment. I see a familiar challenge in the need to keep reaching people and have care ready.
A husband and wife make promises before God that deserve care long after the wedding.
Brands are reworking their websites for AI-assisted shopping. Clear product details can help customers compare, choose and follow through.
Henry faces his former team Sunday at 1 p.m. Eastern. Baltimore needs a home win, and a battered offensive line has a tough job ahead.
Shared Pages let people bring their own agents to the same work. That's a useful next step for teams learning to build with AI.
Instagram's Edits assistant could help explain audience data. I'd use it to find something worth improving in the next video.
OpenAI says it's putting more computing power into safety and watching every training run. I want to see that work succeed.
More companies say AI is helping. That gives us a reason to take our own learning seriously.
Persistent agents and lower model costs have me thinking about the work we could put within reach.
H Company's new models can work through screens, code and software connections. The releases have different licenses, and the results are company-reported.
Piedmont Newnan brings recovery support to the bedside. In our work in Kensington, we see why reaching someone while they are ready for help matters.
PwC's 2026 workforce survey shows uneven access to AI's benefits. For me, learning also means knowing enough to judge the advice I'm being sold.
Shopify's new browser tools let an agent help finish a purchase, with buyer confirmation still required. Here's what changes and where the limits are.
Team Bots share instructions, files and skills across a team. xAI says individual conversations stay private. The feature is now in public beta.
I support athletes getting paid. I don't support making them sit out when they need another chance, or starting their college clock before they get there.
OpenShell sets rules outside an AI agent's control. NVIDIA's separate Sentry design adds hardware monitoring. Here's what those layers are meant to do.
Princess says its new app can compare cruises and check prices inside ChatGPT. It could make the conversation about where to go a lot more useful.
Two centers were injured against Dallas. Rookie guard Vega Ioane stepped in for eight snaps on Baltimore's final touchdown drive. He'd practiced for the possibility.
Lamar Jackson found Zay Flowers, and Tyler Loop hit from 56 yards to beat Dallas. Baltimore had practiced the finish. The defense still needs work.
Publishers are putting more effort into newsletters, subscriptions and their own sites as search changes. That matters to what I'm building here, too.
OpenAI Academy has added courses for developers, leaders, teachers and college students. People also need time to try what they learn.
At Fulton Sheen’s beatification in St. Louis, Cardinal Tagle spoke about the faith and care for people behind his broadcasting career.
Google’s Live Avatar can respond during a conversation. It’s available through Gemini Enterprise, with extra restrictions on custom avatars.
Google reports projects in Uganda and India that use its medical AI models. The work is promising, but the reports don’t establish better care yet.
Faith, business and football all belong on this site. The challenge is to make each article useful and help readers find the subjects they care about.
OpenAI reports that a research agent reached an outside chatbot and an automatic stop failed. The response shows what teams need to test before giving agents more access.
Pepperdine has launched an Institute for Wisdom and AI. I want Christians to learn the technology well enough to help decide how it is built and used.
AI can change how we solve problems without changing what a human being is. Christian ontology places frontier AI beneath human dignity and stewardship.
Microsoft's new Copilot combines Home, Code, and Autopilot. The key questions are ease of use, availability, and the real cost of usage-based agents.
Anthropic’s book-trading experiment raises a practical question about AI delegation: how do we check that an agent understands the person it represents?
Akamai’s Anthropic agreement puts CPU capacity in the spotlight. Running useful AI agents requires infrastructure well beyond the model itself.
Growing interest in baptism in France invites Christians to take spiritual questions seriously and give newcomers the patient attention discipleship requires.
Niswonger DataWorks plans to connect rural Tennessee students with AI projects, college credit, and credentials. The value will depend on the work and support they receive.
Ando puts AI agents inside team conversations. Useful collaboration still needs clear responsibilities, limited access, and human decisions.
Docker Cloud Sandboxes bring long-running coding agents to managed compute. Their usefulness depends on clear scope, reliable environments, and review.
AutonomyAI connects product evidence, planning, and code. Its launch raises a practical question about how small teams preserve judgment across those steps.
A new institute backed by major philanthropic and public partners is proposing a practical standard for AI: test whether it raises incomes, expands agency, and strengthens human flourishing.
Verizon’s new AI-skills initiative recognizes that access to courses is not enough. Trusted community institutions may be the bridge that turns AI availability into economic opportunity.
Meta’s Muse points toward a computing interface that follows the user across devices—and AI glasses make that agent present in the physical world.
An international home game carries a real cost for Baltimore. It can also deepen the Ravens’ audience and help the NFL grow, provided the organization treats expansion as a lasting relationship rather than a one-week event.
YouTube is putting research, editing, testing, channel management, and business support into one creative environment. That gives an independent creator remarkable capacity, but it also makes a recognizable point of view more valuable.
Persistent memory and recurring workflows can give a small seller operating capacity that once required a larger team. The gain is real, but only if the owner defines the boundaries and remains responsible for the result.
The firm is creating a place to build new AI businesses at the edge while spreading useful AI capability through the core. The structure matters because adoption is not the same thing as putting every tool in every employee’s hands.
Cadence’s ChipStack system reaches into demanding design and verification work where context and correctness matter. Its real promise is not autonomous engineering; it is a better division of labor between technical experts and machines.
Baltimore’s latest blown lead belongs to a long-running pattern. Jesse Minter didn’t create it, but it is now his responsibility to change it.
APA’s 2026 Work in America survey shows AI use rising alongside worker concern. Kris McFadden argues that leaders owe people explanation, preparation, shared language, and visible ownership.
AI often shifts work into human validation, correction, and contextual judgment. Leaders need shared language and training that makes that work visible.