PWV Portfolio Feed

🚀 We just shipped v0.230!

Screen sharing in Zed collab now works on Wayland/Linux. 🐧🖥️

74 Netlify sites supporting global presence. ~70 active projects across brands and internal services.
Zero emergency calls.

That last one is the one Mark actually cares about.

Serverless functions across 18 sites. Built-in build and edge capabilities.

Dynamic use cases handled without adding infrastructure overhead. Developers focused on React, not EC2.

Netlify wasn't a one-off adoption. It became the core digital platform.

Git + CI-driven deploys as the default. Teams moving faster without the multi-hurdle process of before.

The goal: consolidate into e-comm and non-e-comm stacks, and let full-stack developers actually own delivery.

Simple in theory. Hard when your infrastructure is fighting you.

Even basic fixes required sysadmin, cloud admin, network admin, AND DevOps.

Every. Single. Time.

That's not an ops problem. That's a platform problem.

Properties scattered across their own servers and multiple clouds.

Problems cropping up around the clock. No clean way to own any of it.

Barbie. Hot Wheels. Fisher-Price. American Girl.

That's just four. Mattel had 500+ digital properties to manage, and their stack wasn't built for that kind of scale.

500+ digital properties. Issues day and night. Every fix pulling in sysadmin, cloud admin, network admin, and DevOps.

Mattel called it "death by a thousand cuts."

Here's how they got out. 🧵

Effective immediately, we are making the vite-plus repository private.
The source will be distributed exclusively through our production sourcemaps.

We believe this is the only logical path forward given the recent events.

This does not change our stance on Open Source!

AI use case: Task planning

AI can break complex coding tasks into clear, executable steps. It helps you move faster by turning ambiguity into structured action.

i'm not able to get the val version to work fyi - lmk if you'd like a hand with it

means a lot coming from Sam!

:)

In Prisma Next, streaming will feel just like writing plain TypeScript rather than a separate framework feature.

You'll simply build the query, execute the plan and for await the rows.

AI Terminology #22: Logits
↳ The raw, unnormalized output values from a model before probabilities are applied, typically passed through a softmax function.