"Battered by years of mass layoffs, California tech workers were hoping the job market would rebound this year," reports the Los Angeles Times. "But things are getting worse." The class divide is widening in Silicon Valley as a tiny group of employees is landing unprecedented packages for AI skills, while many others struggle to find work. The have-nots are doing everything that used to guarantee great jobs — refreshing resumes, optimizing LinkedIn profiles and doing interviews — but companies are much more picky these days. The tech jobless are rethinking their lives. Some are taking pay cuts, others are leaving tech. Some are going back to study or launch startups. Some have retired.... Since 2022, more than 815,500 tech workers have been laid off, according to Layoffs.fyi, a website that tracks job cuts. The tsunami of pink slips surged in 2023, when companies that had gone on hiring sprees during the COVID-19 pandemic began to cut back. From January to April, U.S. tech employers announced 85,411 job cuts this year, up 33% from the same period last year, according to global outplacement and executive coaching firm Challenger, Gray & Christmas. The Public Policy Institute of California estimates that the number of information jobs — which includes jobs in hard-hit Hollywood as well as tech — tumbled 17% between the middle of 2022 and this February. The San Francisco Bay Area has been hardest hit, the institute said in a recent report, with the number of jobs declining by 0.4%, compared with 7.5% growth over a similar time span before COVID-19 slammed into the U.S. economy. Tech layoffs are also spilling over into other industries. Automaker General Motors laid off roughly 600 workers in its information technology department, and Walmart is reportedly laying off or relocating roughly 1,000 workers in its technology and products teams. Recruiters say companies have become much more selective, requiring AI skills, combining different positions and interviewing more people for each job. "You're seeing elongated hiring cycles," said Robert Lucido, senior director of strategic advisory at Magnit, a California company that helps tech giants and other businesses manage contractors, freelancers and other contingent workers. "There's more opportunity to fill the need that they truly want." Paul Flaharty, district president at staffing firm Robert Half in Los Angeles, said companies are laying off workers, but also creating new roles tied to AI initiatives. "For individuals that are displaced, it's really important that they find ways to upskill themselves so that they can make themselves as attractive as possible for these new jobs that are being created," he said. Kira Martins was already taking on more work in a small team at Snap — the parent company of disappearing messaging app Snapchat — when she was laid off in April. The company said the layoffs were to cut costs as it focuses on profitability, noting how employees are using AI to "reduce repetitive work, increase velocity, and better support our community, partners, and advertisers...." Martins, a 36-year-old Los Angeles resident, views AI as a tool and is optimistic about finding her next role. People still need to decide how to use AI and check the work it generates, she said. "In tech, you want to be a first adopter, because if you don't move quickly, it's very easy to become irrelevant," she said. "Everyone's kind of hopping on the AI train." A former Google worker (laid off more than a year ago) says he's still job hunting, according to the article, and "he's learned it's not enough to just apply in this competitive market. Workers really need to network and leverage their connections to get seen by hiring managers and stand out." But when 64-year-old product manager Bruce Bowers lost his job at Oracle — along with thousands of others — he just started his retirement early.

Read more of this story at Slashdot.

Microsoft’s Outlook for Windows has a notification problem that is hard to ignore. Clicking a Windows 11 notification for a new email is supposed to take you straight to that message. Instead, the new Outlook makes you wait, and the numbers are embarrassing.

Windows 11 ships with two versions of Outlook. There is Outlook Classic, the long-running Win32 desktop app built for power users, and there is the new Outlook, which Microsoft is pushing as the future of email on Windows. The newer one is built on WebView2 and is, in essence, a browser window that loads Outlook.com. If you have ever used both side by side, you already know which one feels faster and which one does not.

Outlook has had a complicated reputation for years. The original Win32 app became infamous for being bloated and difficult to configure. Microsoft’s answer was to ditch native code and rebuild from the web up. The result, called the new Outlook, replaced the lightweight UWP Mail and Calendar apps that some Windows users had grown used to. Windows Latest reported back in 2023 how users protested when Microsoft announced plans to retire those UWP apps in favor of a web wrapper. The company pushed ahead anyway, and by late 2024, the Mail and Calendar apps were officially shut down.

Microsoft has also been pushing the new Outlook at enterprises, though it postponed the forced opt-out deadline to March 2027 from the originally planned April 2026. A delay of a full year shows that even Microsoft knows the app is not fully ready for every workload. New Outlook has improved in real ways since launch, but the performance story is still a mixed one, and nowhere is that more apparent than in how it handles notifications.

New Outlook takes 10 seconds to go from notification to the respective mail

Before getting to the frustrating part, credit where it’s due, New Outlook used to be noticeably slow to launch from scratch, but not anymore.

Outlook (classic) vs New Outlook:

New Outlook now opens almost as fast as Outlook Classic, which is still slightly quicker of the two. But I would say both are neck and neck, at least when it comes to opening speeds.

However, when a new email arrives in Windows 11, a notification banner appears at the bottom right of your screen, and that’s where the problem starts. Clicking that banner, or from the Notification Center, is supposed to take you directly to the email that triggered it.

With Outlook Classic, it opens that specific email almost instantly.

With the new Outlook, clicking the notification opens the app, loads the full inbox, and then takes around 10 seconds before the specific email from the notification shows up on screen.

What makes this even more absurd is that if you ignore the notification banner and instead open Outlook directly from the Start menu, you can find and click the new email from within the app and be done with it, all before the notification banner even disappears from the screen.

Five seconds to open Outlook and click an email manually. Ten seconds wait time to see that same email if we click the notification directly. This is ridiculous, even for Microsoft.

And, as it turns out, this isn’t a problem Microsoft can easily fix with an app update.

Outlook is based on WebView2, and like all web apps, it’s slow

New Outlook is built on Microsoft Edge’s WebView2 runtime, which is a Chromium-based rendering engine. Every time you interact with the app, including clicking a notification, a browser-like process chain has to do the work. The app has to initialize or resume its web layer, authenticate, load the relevant mail thread, and render it, all through that web engine.

As Windows Latest reported in December 2025, Microsoft acknowledged this slowness and was testing a new API called “Delayed Message Timing” to help diagnose performance issues in WebView2 apps. However, we haven’t seen any use of that API while clicking Outlook notifications.

New Outlook runs as 10 separate processes in Task Manager, compared to Outlook Classic, which runs as a single compact process. The list inside the new Outlook includes WebView2 Manager, multiple WebView2 Utility processes, a WebView2 GPU Process, a WebView2 Service Worker, and more. Each of those is essentially a browser component. They all consume memory individually, and they all take time to resume from a suspended state when you click a notification.

Speaking of memory, the new Outlook uses between 490 MB and 636 MB of RAM while idle, with individual sessions varying based on mailbox size. Outlook Classic, doing the same job, uses around 117 MB to 148 MB at idle. A roughly fourfold difference.

As for CPU, new Outlook uses around 4% at idle while Outlook Classic uses less than 1%. These numbers are from my own measurements using Task Manager with both apps open simultaneously.

Of course, these issues are common to all web apps. As we reported, WhatsApp now consumes 1.2 GB of RAM doing nothing after Meta replaced its native WinUI app with a WebView2 wrapper.

Microsoft has been aware of the offline and performance limitations of new Outlook for some time. The company spent much of 2024 trying to make the app work properly without an internet connection, something Outlook Classic handles natively by caching mail locally. A web app, by design, is always reaching out to a server.

New Outlook is improving, but the gap with Classic isn’t closing anytime soon

In fairness, the new Outlook has come a long way since its rocky debut. The March 2026 update added better folder search options and improved shared mailbox access, two areas where the app lagged behind Classic for a long time. The May 2026 update brought automapped calendar support, so switching from Classic to new Outlook no longer drops your shared calendars. Teammate calendars now show up automatically in the navigation pane.

More recently, Microsoft confirmed a June 2026 update with five notable additions, including an all-accounts inbox view (also called Unified Inbox) arriving in August 2026, improved mail merge, and expanded .PST support. The .PST import update in July 2026 will let users bring in calendar items and contacts from local archive files, which was a long-standing pain point for anyone switching from Classic.

The push to get people to switch is getting louder too. Microsoft listed 15 productivity features in early June 2026 as reasons to make the move from Classic. The list includes offline access, richer Copilot integration, faster search, improved calendar controls, and more. Many of those features are things Classic users have had for years, which makes the framing a bit odd, but the direction is clear. Microsoft wants new Outlook to become the default experience for everyone.

We were also told in late 2025 that a calendar agenda view in the Notification Center was coming, bringing back a Windows 10 feature that went missing with Windows 11. The agenda view, when it arrives, will also be powered by WebView2. Whether it introduces similar delays remains to be seen.

A web app cannot fix performance issues

Microsoft celebrated growth for new Outlook in 2024, but a significant portion of that growth came from forced migration. People did not choose the web app because it was faster. They were moved to it because the apps they previously used (Mail + Calendar) were shut down.

New Outlook opening fast from the Start menu is a real improvement. The work being done on new features shows that Microsoft is listening to complaints. But until the notification experience matches what Outlook Classic has been doing without issue for years, the new Outlook is still working around a fundamental constraint imposed by the WebView2 architecture.

The only solution, as you might’ve guessed, is the move to WinUI. We already reported that Microsoft is now fully committed to WinUI, with Rudy Huyn preparing a team to make native Windows apps, and so we may see a native Outlook too…

For now, if fast notification handling is important to your workflow, Outlook Classic is the more reliable choice. Classic Outlook is still available to download and is supported until April 2029. The new Outlook will keep improving, but some of its limitations are baked into how it is built, and those are harder to fix with a feature update.

The post Microsoft’s new Outlook takes 10 seconds to do what Outlook Classic does instantly on Windows appeared first on Windows Latest

What makes this combustible: at the very moment that tens of thousands of workers are being shown the door, a small cohort of AI insiders is becoming wealthy on a scale that's hard to comprehend.

The UK is the latest country to follow Australia in implementing a total social media ban for children under 16, Prime Minister Keir Starmer has announced. The ban, which could take effect from early next year, will be joined by wider measures that will also prevent children from talking to strangers in online games, livestreaming, or using sexual or romantic chatbots.

"Do we truly believe that social media creates a happy environment for our children? Do we truly believe that it's a place where they can feel safe?" Starmer asked during a press conference announcing the policy. "I don't think I even need to answer those questions, do I?"

" …

Read the full story at The Verge.

This is a bit of rambling from me and what I believe is a good setup for developing software together with AI tools. I believe the AI tools are good, which will help good developers produce better software for our end clients.

What is the aim of creating software?

This is a super hard question because it is not always the same for different dev setups, but at some point, in the production of the software and the company paying the bill, the aim is to produce as much value as possible for the least amount of cost and within the time requirements. The least amount of cost is for the full lifecycle and not just the creation of the software.

How does AI fit, in the future development processes?

AI will be a large annual cost for the software development process. Software needs to be paid for with value. At present, the companies providing these AI services are not making profits and so the AI costs must go up. This means that if we use AI to produce software, the costs must be covered. This will only work if we become more efficient. Even the companies which are leading the way in software development with AI are not meeting the required cost targets once the price goes up. The hope is that the tools will get better.

Who can use AI efficiently?

This is actually really hard to answer and not clear. A lot of people proclaiming more speed, and amazing solutions are not really being honest. A big problem is, to use AI efficiently, you need to be a domain expert in the area where you use AI. So, if I use AI to produce security code, I can be faster because I can judge, if the output is good or bad. If I use AI somewhere where I do not understand the output, I will produce a worse solution than if I did not use AI. This is because without AI, I would read it, learn, ask experts, and educate myself, what is good in this domain. There are still no short cuts to this process for producing production code.

The skills we need in the future are people who understand their domains. Someone that can code good, will be able to code with AI. Someone who is not so skilled will produce a high amount of slop and slow down the whole team or reduce the quality of the product.

What do we need as software developers in the future?

One of the biggest challenges we have now is finding access to real, reliable and quality information. The internet is getting filled with AI slop and the people producing quality software blogs are declining. Stack overflow seems to be used less. Less blogs are being created because there are no rewards anymore. The content gets taken by AI bots and shared without any recognition. The payment, reward models are broken. People with knowledge or access to real knowledge will be key in the future.

What type of dev teams do we need in the future?

We need domain experts. And we need a way to train people to become domain experts. When hiring, people who learn to understand the topics are the skilled professionals we need and not the ones who are good at prompting. I think future successful dev teams will be small teams with very strong developers who can talk to the client and understand the domain. Funny thing, this was the same before AI when quality and costs are the main drivers.

What about outsourcing?

If AI brings all the promises it gives, this industry will be required less in the future, because I can just use AI to implement the features. The engineering work is what is still required. So code experts, architects, domain experts, these are the skills which will be still required. People close to the client, people who speak the same language are the future.

How will this affect project team setups?

We need more senior technical people and domain experts and less medium people. Good teams will be smaller and closer to the client. Less agile processes and less product team management is required. Closer to the client with experts is the key. This would require a complete revamp of how the industry does and creates software.

What about debugging and monitoring?

This is one of the areas where AI can shine, if the applications are created with quality. If the right information and the correct logs are created using a good tool, AI can be used to find all sorts of operational or performance issues. This will depend on the quality of the application but this is an area with loads of potential for efficiency gains.

Should we let AI complete PRs?

Absolutely not. We are responsible for the code, and at the center of every agent, or AI process, is a non-deterministic piece of software. This will choose a probable answer or anything that will fulfil the prompt request. It has no intelligence, just probability and statistical decision-making. To produce maintainable software, the dev team must understand this, otherwise the quality will suffer. A person is required between the deterministic conversions and the non-deterministic AI parts. This is why we do not need to understand assembly, but we do need to understand the code. C# to assembly compiles and always returns the same.

AI and security

This is the bit which worries me the most. AI will execute any instruction it is given. It does not think. If AI tools have access to all your data, there is a possibility that your data is shared with services which should not get your data. If you let AI act on your behalf, this is even more dangerous and the best answer for the prompt is not always what you want. GDPR, data protection and client NDA agreements are regularly getting broken when using AI in software processes. There are some great guidelines on security from OWASP and this is something I need to invest in.

AI and the planet

When we use AI, we use a large amount of energy and water, and we are no longer working in a clean industry. I think at some stage, the energy factor should also be paid for and must be visible. We need to understand how much energy and water was used to create the feature X. If I know what I use, then I can make a decision, if this was worthwhile or not. At present, this is not transparent.

Which AI tools do I use

Almost all of them in the Microsoft world. I enjoy Visual Studio Copilot and Visual Studio Code Copilot using different models depends on which delivers the best results. I really like the Github copilot.

Artificial Intelligence is rapidly evolving beyond traditional chatbots. Today, developers are building intelligent systems where multiple AI agents collaborate, retrieve knowledge, and solve problems together. Microsoft's Agents League Hackathon provided the perfect opportunity to explore this new approach through the Reasoning Agents challenge.

For this challenge, I built ShadowQuest, a fantasy role-playing game (RPG) powered by Microsoft Foundry, Foundry IQ, Azure AI Search, GPT-4.1, and GitHub Copilot. The project demonstrates how specialized AI agents can work together while using Retrieval-Augmented Generation (RAG) to deliver accurate and context-aware responses.

Microsoft Agents League is a global developer challenge designed to encourage developers to build intelligent AI applications using Microsoft's latest AI technologies. Participants could choose from three tracks: Creative Apps, Reasoning Agents, and Enterprise Agents.

I selected the Reasoning Agents track because I wanted to explore how multiple AI agents could collaborate instead of relying on a single large language model. Another important requirement for this year's challenge was integrating at least one Microsoft Intelligence Layer. For ShadowQuest, I chose Foundry IQ as the project's intelligence layer.

 

Fantasy RPGs are built around storytelling, exploration, and collaboration between different characters. Every character usually has a unique role, whether it's a warrior protecting the team, a mage interpreting magical knowledge, or a rogue discovering hidden paths.

I wanted to recreate this experience using AI.

Instead of building one AI assistant responsible for everything, I designed a system where multiple specialized agents collaborate to create a richer and more immersive adventure.

ShadowQuest is set in a fantasy world filled with magical artifacts, forgotten kingdoms, mysterious locations, and story-driven quests. Players can ask questions about the world, explore different locations, and learn about the game's lore through conversations with AI agents.

The architecture follows a simple but scalable design.

At the center of the system is the Game Master Agent, which acts as the orchestrator. Every player interaction starts with the Game Master. It receives the player's request, determines what information is needed, retrieves additional knowledge when required, and generates the final response.

Supporting the Game Master are three specialized agents:

• Warrior Agent – Focuses on combat strategy and tactical decisions.
• Mage Agent – Provides magical knowledge, world lore, and information about ancient artifacts.
• Rogue Agent – Specializes in exploration, investigation, and discovering hidden information.

Each agent has a clearly defined responsibility, making the system easier to understand, maintain, and extend in the future.

 

 

 

One of the most important parts of the project was integrating Foundry IQ.

Instead of storing every piece of game information inside prompts, I created a dedicated knowledge base containing information about characters, magical artifacts, locations, quests, and the history of the ShadowQuest world.

This approach separates knowledge from reasoning.

Whenever a player asks a question, the Game Master Agent first retrieves relevant information from the knowledge base before generating a response. This ensures that answers remain consistent with the game's world while reducing hallucinations.

Foundry IQ became the central source of truth for the entire project, making it easy to manage and expand the game world without constantly modifying prompts.

 

To enable intelligent retrieval, I connected Foundry IQ with Azure AI Search.

The RPG documents were indexed, and vector embeddings were generated using Microsoft's embedding models. This enables semantic search, allowing the system to understand the meaning behind a player's question instead of relying only on keyword matching.

For example, if a player asks about a magical relic without mentioning its exact name, Azure AI Search can still retrieve the correct information based on semantic similarity.

The complete workflow looks like this:

1. The player submits a question.
2. The Game Master Agent receives the request.
3. Foundry IQ queries Azure AI Search.
4. Relevant documents are retrieved.
5. GPT-4.1 generates a grounded response using the retrieved context.

This Retrieval-Augmented Generation (RAG) approach significantly improves the quality and reliability of responses.

 

GitHub Copilot played an important role throughout the development process.

It helped generate Python classes, improve documentation, create helper functions, and speed up repetitive coding tasks.

During the live demonstration, I also showed how Copilot could quickly generate a new Healer Agent, demonstrating how AI-assisted development makes it easier to extend a multi-agent application while maintaining a consistent architecture.

Rather than replacing the developer, Copilot acted as an intelligent coding assistant, allowing me to focus more on architecture and design decisions.

 

During the Microsoft Agents League Reasoning Agents Battle, I demonstrated the Game Master Agent by asking questions about the ShadowQuest world, magical artifacts, and game lore.

One of the most interesting parts of the demonstration was observing the retrieval process.

Before generating a response, the Game Master Agent called the knowledge retrieval function through Foundry IQ. This confirmed that the system was retrieving relevant information from the indexed knowledge base rather than relying only on GPT-4.1's internal knowledge.

This demonstrated how RAG can create more grounded, reliable, and context-aware AI experiences.

 

Building ShadowQuest taught me that designing multi-agent systems is as much about architecture as it is about AI models.

Clearly defining responsibilities for each agent made the application easier to maintain and opened the door for future expansion.

I also learned how valuable Retrieval-Augmented Generation can be for applications that depend on structured knowledge. Separating reasoning from knowledge allows AI systems to remain accurate while making it easier to update information over time.

Finally, participating in the Microsoft Agents League was an incredible opportunity to experiment with Microsoft's latest AI technologies, learn from other developers, and share ideas with a global community passionate about agentic AI.

ShadowQuest is only the beginning.

In future iterations, I plan to expand the project by introducing additional agents such as a Merchant Agent and Healer Agent, implementing persistent player memory, adding dynamic quest generation, improving combat mechanics, and enabling deeper collaboration between agents.

These improvements will make the game world more immersive while continuing to explore the possibilities of agent-based AI systems.

ShadowQuest demonstrates how Microsoft Foundry, Foundry IQ, Azure AI Search, GPT-4.1, and GitHub Copilot can be combined to build intelligent multi-agent applications.

More importantly, the project reinforced an important idea: the future of AI is not a single assistant performing every task, but a team of specialized agents collaborating with shared knowledge to solve increasingly complex problems.

Participating in the Microsoft Agents League was an inspiring experience that allowed me to explore the next generation of AI development while building a project that combines storytelling, reasoning, and knowledge retrieval. I look forward to continuing this journey and discovering new ways to build intelligent applications using Microsoft's growing AI ecosystem.