Exploring AI that respects your data.
A blog about artificial intelligence and digital privacy — written for people who want to understand both, use capable AI tools, and stay in control of their data. Posts here are practical and grounded: what is actually being built, what it means in use, and what is worth paying attention to.
-
Testing Claude’s New Fable 5 on a Real Infrastructure Problem
When Anthropic released Claude Fable 5 in June, its first Mythos-class model made generally available, I had a task sitting on the shelf that felt like a fair test: a phone-to-AI-agent SIP bridge I had started, stalled, and restarted with earlier model generations more than once over the past few months. It never quite got…
-
Our Voice Agents Can Now Answer Calls
Phone calls have always been the one channel bacher-ai’s voice agents couldn’t reach — everything ran through the web app or the LiveKit Playground. That’s changed. A new SIP bridge now connects real phone numbers directly into the platform’s LiveKit voice agents, and the first agent built for it is already live: a phone-answering assistant…
-
LiveKit Is Now Live on bacher-ai
Real-time voice AI is no longer a cloud-only game. The bacher-ai platform has just deployed a self-hosted LiveKit instance — a significant infrastructure milestone that opens the door to a new generation of voice-driven AI experiences, all running under full data sovereignty. What LiveKit Is LiveKit is an open-source, WebRTC-based platform for building real-time audio…
-
Self-Hosting Your Git Repos with Forgejo
Most developers don’t think twice about pushing their code to GitHub. It’s convenient, it’s fast, and everyone’s on it. But convenience always comes at a price — and with code repositories, that price is handing over your intellectual property to a platform you don’t control. We recently completed the migration of our own repositories to…
-
Preparing for the Quantum Threat to Encryption
Quantum computers do not need to exist at full scale today to be a threat to your privacy today. That is the uncomfortable reality driving one of the most significant shifts in cryptography in decades. Governments, intelligence agencies, and well-funded adversaries are already collecting encrypted internet traffic — emails, VPN sessions, cloud data — with…
-
Run AI on Your Computer, Privately
Every time you type a question into ChatGPT or Claude, that conversation travels over the internet to a distant server — and stays there. It may be read by employees, used to train future AI models, or stored indefinitely. For most casual queries that might feel harmless. But when you ask an AI to help…
-
Protecting Kids Online Without ID Checks
Online age verification is having a moment. Governments around the world are pushing platforms to check users’ ages before granting access to adult content, social media, and other restricted services. The idea sounds reasonable — keep children away from harmful material. But the way most age verification systems work today creates a new set of…
-
OpenClaw — The AI Agent Everyone Is Talking About
In early 2026, a small open-source project called OpenClaw went from niche developer experiment to one of the most-discussed tools in the AI space — accumulating over 250,000 GitHub stars in roughly 60 days, outpacing React’s entire decade of growth. This post breaks down what it actually is and why it matters. A Gateway, Not…
-
Privacy Can’t Wait: Phil Zimmermann Was Right
In the early 1990s, a software engineer from New Jersey released a free encryption tool that the US government classified as a weapon. The tool was Pretty Good Privacy — PGP — and it would go on to become the most widely used email encryption software in the world. The engineer was Phil Zimmermann. He…
Quote