The ampersand is nearly 2,000 years old. yet designers still can’t agree on what it should look like. From ancient Roman *et* to hundreds of wildly different & designs, the story behind typography’s favorite squiggle is stranger than you might expect.

A couple of weeks ago, I talked about the “Happy Path Trap”—how LLMs create massive Day 1 velocity while burying teams in silent Day 2 architectural debt.

During that discussion, I pitched an idea: What if we required an Assumptions & Constraints Manifest (ACM) in every AI-assisted PR to make hidden trade-offs visible?

Since I’m not one to wait around for someone else to build tooling I want to use, I spent the last few days building it out.

Today, I’m open-sourcing ACM v1.1—a formal spec and developer toolkit designed to audit architectural boundaries, not just syntax.

What’s in the toolkit:
Typed Contracts & Proof Obligations: Requires the PR to explicitly state guarantees around concurrency, idempotency, and scaling—citing exact line numbers and locking mechanisms.

Automated Generation: Drop-in .cursorrules and PR templates that prompt Cursor, Claude, and Copilot to generate manifests automatically during code creation.

CI/CD Linter: An open-source CLI linter (@roblarsen/acm-cli) and GitHub Action to validate manifests and gate PRs automatically in your pipeline.

Instead of spending 45 minutes playing digital archaeology on AI-generated diffs, reviewers can audit explicit boundaries in 30 seconds.

Get Started
GitHub: github.com/roblarsen/ACM

CLI: npx @roblarsen/acm-cli

Hundreds of thousands of AI agent sandboxes can already run concurrently on a single DeepSeek cluster. Now the company is staffing up to handle what happens as that number — along with its training, evaluation, and other backend workloads — keeps climbing.

Cui Tianyi, who joined DeepSeek in March and works on its Harness team, the group responsible for the infrastructure and environments used to run and evaluate agents, announced in an X post that roughly 150 engineering positions on September 7, with the hiring concentrated in server-side engineering and Agent Elastic Compute rather than AI research. The work spans operating systems, virtualization, networking, storage, scheduling, and the control-plane services that coordinate those resources.

Cui said DeepSeek’s existing backend systems will need upgrades, maintenance, and rewrites as workloads grow. One such system at the center of that scaling challenge is DeepSeek Elastic Compute, or DSec, the sandbox infrastructure DeepSeek built to execute agent workloads during post-training and evaluation.

Cui said DeepSeek’s existing backend systems will need upgrades, maintenance, and rewrites as workloads grow.

Four sandboxes, one SDK

Agent workloads require more than GPUs for inference, with each agent also needing an isolated environment to run code, call tools, change files, and collect the results.

DSec supports four types of those environments through the same Python SDK. Simple function calls go to pre-warmed containers, while Docker-compatible containers handle jobs that need a persistent environment. DeepSeek uses Firecracker microVMs when stronger isolation is needed and QEMU virtual machines for workloads that require a full guest operating system.

That range means the same infrastructure can handle anything from a simple tool call to a software-engineering task that needs an entire OS. It’s a similar challenge to the one the rest of the industry is bumping into as agents move from demos to production. OpenAI, for instance, recently designed custom silicon specifically to address the compute pressure that agent workloads create, and DeepSeek open sourced its own agent harness in August.

Lazy loading agent environments

Every sandbox needs its own environment, but copying complete container or VM images onto every host would consume enormous amounts of storage and network bandwidth while adding to startup time. DeepSeek gets around that by tying DSec into 3FS, the distributed filesystem it originally built for its AI infrastructure, and keeping container base images and filesystem commits as read-only layers backed by 3FS.

The metadata stays local, but the underlying data blocks are fetched only when they’re actually needed. MicroVMs use a similar setup, sharing their read-only base layer through 3FS while writes from individual sandboxes are kept in local copy-on-write layers.

DeepSeek says DSec reduces duplicate page-cache usage across virtualized environments and reclaims memory to allow safe overcommitment, while changes to the container runtime cut the CPU overhead of each sandbox.

The team also had to deal with spinlock contention inside the container runtime. At small scale, the CPU time spent there barely registers. At scale, it limits how densely those environments can be packed onto each host.

DeepSeek says DSec reduces duplicate page-cache usage across virtualized environments and reclaims memory to allow safe overcommitment, while changes to the container runtime cut the CPU overhead of each sandbox.

When replay breaks training

During reinforcement learning and other post-training workloads, large numbers of agent rollouts can be running at once, and jobs may be interrupted as compute gets reassigned. Starting over wastes everything the agent has already done, but picking up where it left off isn’t as simple as replaying its previous commands.

Some of those commands may have changed a file or otherwise altered the environment, so running them again could produce a different result or leave the training trajectory in the wrong state. DSec avoids that with a globally ordered trajectory log that records commands along with their results.

When a rollout resumes, DSec can fast-forward through the completed work using those recorded results rather than executing the commands a second time. That reduces the cost of interruptions across thousands of training and evaluation runs, while the same logs preserve a history of how each sandbox changed and allow earlier sessions to be replayed.

Engineers, not researchers, wanted

The roughly 150 openings reach across DeepSeek’s backend, including the lower-level systems work behind Agent Elastic Compute as well as the services that support its models and agents.

DeepSeek said in June that it planned to at least double the size of every department, but this round of hiring leans heavily toward the systems underneath its models rather than the models themselves. DSec is part of that work, with hundreds of thousands of sandboxes running concurrently and putting pressure on everything from how jobs are scheduled to how they recover after an interruption.

The roughly 150 openings reach across DeepSeek’s backend, including the lower-level systems work behind Agent Elastic Compute as well as the services that support its models and agents.

The post DeepSeek is hiring 150 engineers, and none of them will touch a model appeared first on The New Stack.

Meta has launched Muse, a personal AI agent developed under chief AI officer Alexandr Wang. "The product, long in development, was touted as a key next step by CEO Mark Zuckerberg in his recent 6,500-word manifesto," reports Axios. From the report: Muse, as the agent is known, exists in a chat interface, similar to a text thread. It's designed to be more proactive and long-running than typical chatbots. Users can name their agent, create an avatar and customize how it communicates. The Muse agent runs on a dedicated virtual machine in Meta's cloud, using a built-in browser that's visible to the user. Meta is offering a free tier of Muse, as well as two subscription options, at $20 per month and $100 per month. "For the vast majority of users, they should be able to do what they need to within the free tier," Wang told Axios. "But for real power users, you know, those subscription tiers help us cover the computer costs." There is no advertising within Muse, but Wang said the company is exploring commerce opportunities that could generate additional revenue. Initially Muse will be available in the U.S. and works on iOS, Android and the web, with support coming soon for Meta's AI glasses. "The full vision in the future is we want to develop personal superintelligence that helps people accomplish their goals, pursue their passions, build things that they never would have built if they didn't have the technology," Wang told Axios. Meta offers users more privacy controls with Muse than in its previous AI products, including the option to prevent queries from being used by Meta and a planned confidential mode where the company cannot see activity inside a user's virtual workspace. There's also an entirely separate system called Sentinel that governs Muse's access to the internet and connected services.

Read more of this story at Slashdot.

OpenAI announced ChatGPT Images 2.5 on Tuesday and is adding a new way to tell ChatGPT what you want it to make an image of: by drawing a doodle. With a new feature called Sketch, you can just draw something right inside ChatGPT and then tell ChatGPT how you want it to make an image based on that sketch.

You can activate the Sketch feature by typing @Sketch into your chat box, and a window will pop up where you can make your drawing. I briefly tested it, and it successfully turned my bad doodle of a cat that I drew with my computer mouse into a realistic photo, per my instructions. You can also now leave comments directly on parts of images …

Read the full story at The Verge.

We are pleased to announce the release of .NET 11 Release Candidate 1. Follow the main announcement and blog announcement.

This is our first of two release candidates which come with a go-live support license so you can confidently use this release for your production applications. This release of .NET 10 is supported in the new Visual Studio 2026 Insiders release alongside Visual Studio Code with the C# Dev Kit. Check out the full release notes below and get started today.

This release contains the following highlights and improvements.

Browse the release notes for a full list of improvements in this release:

• .NET MAUI
• ASP.NET Core
• C#
• F#
• Container images
• EF Core & Data
• Libraries
• MSBuild
• NuGet
• Runtime
• SDK
• Windows Forms

Get started

To get started with .NET 11, install the .NET 11 SDK.

If you're on Windows using Visual Studio, we recommend installing the latest Visual Studio 2026 Insiders. You can also use Visual Studio Code and the C# Dev Kit extension with .NET 11.