Engineering notes on distributed inference

Blog header reading “Using Apple’s Local Models as Claude Subagents,” next to a cartoon apple wearing a chef’s hat and apron on a stacked card labeled “On-device.”

Use Apple Foundation Models as Claude Code Subagents

Announcement

yeschef now supports Apple Foundation Models on macOS 27, so Claude Code can delegate smaller tasks to Apple’s on-device model and save on your Claude usage.

Read more
Blog header for Cascadia titled “Running Local AI on Business Laptops: What Hardware Do You Actually Need?” On the left, white text and the Cascadia logo sit on a black background. On the right, a person types on a laptop in a dimly lit room, and the screen shows a dark terminal full of code and system commands.

Research

Running Local AI on Business Laptops: What Hardware Do You Actually Need?

Many business laptops and workstations can now run capable AI models locally. This post covers what running AI on-device actually involves and how much RAM you need for different kinds of work.

Read more
Promotional graphic announcing “Cascadia now supports Inkling.” The image shows the Inkling AI model by Thinking Machines at the top, connected by branching lines to four Intel AI PCs below. Cascadia branding and logo appear in the lower-left corner.

Research

Running Inkling (975B) Across Eleven Intel PCs with Cascadia

We ran Thinking Machines' Inkling (975B) across eleven Intel Core Ultra PCs using Cascadia, reaching a peak of 60.29 tok/s of aggregate decode throughput with 88 concurrent requests.

Read more
Blog thumbnail titled "Where NPUs Fit into LLM Inference" with the Cascadia logo below it. On the right, a glowing chip labeled NPU sits at the center of a dark circuit-board grid.

Research

Where NPUs Fit into LLM Inference

Most modern laptops now ship with an NPU, but few people use it for running LLMs.

Read more
Cascadia banner reading 'Building a permissioned AI inference network,' beside a nighttime photo of an office tower's grid of lit windows.

Research

Building a Permissioned, Peer-to-Peer AI Inference Network

How a permissioned peer-to-peer system can turn a fleet of PCs into a single, on-prem inference endpoint.

Read more

Latest blog posts

Bring a fleet. Keep everything.

We stand up your first private agent in under a week, on your hardware, on your network. You keep all of it.

Free for individuals

Download

Apply to become a design partner

Request a Pilot