Blog

Notes on what we're building and why.

  1. What does it cost to run a Kimi K3 at 100 t/s at 1M context window? [an exploration]

    LinkedIn

    An exploration of what it really takes to serve a frontier model like Kimi K3 at 100 tokens a second with a 1M-token context.

  2. What's Wrong with Alexa & Google Home in 2026

    Deep Research

    A look at the biggest frustrations users face with today's smart speakers — from privacy concerns to reliability failures.

    Generated using ChatGPT Deep Research — linked to the original conversation for transparency.

  3. How Piccolo Made Me Fall Back in Love with Math

    LinkedIn

    The elegant math behind Piccolo's zero-trust storage: tiny Merkle proofs that keep a federation honest.

  4. The Digital Factory Seal: How Devices Prove They're Genuine

    LinkedIn

    A short read on how we use cryptography to ensure that every Piccolo device is authentic.