Descriptions:
Fahd Mirza walks through installing and configuring NullClaw, a fully autonomous AI assistant written in Zig — a low-level systems programming language designed for performance and minimal binaries. Part of the OpenClaw family of tools, NullClaw compiles to a single 678-kilobyte binary that consumes approximately 1 MB of RAM and boots in under 2 milliseconds, making it the smallest and fastest member of the family.
The video covers building NullClaw from source on Ubuntu, including resolving a missing SQLite3 library dependency that causes a build error, then running the onboarding wizard to configure a locally running Ollama model as the AI backend. Mirza explains the configuration options including supervised vs. fully autonomous operation modes and SQLite-based memory storage. He then demonstrates how to create a Telegram bot via BotFather, obtain a bot token, and insert it into NullClaw’s config.json to enable Telegram as the messaging interface.
The end result is a privacy-respecting local AI agent — no cloud API required — accessible through Telegram on any device, with the entire inference stack running on local hardware. All commands and the config template are linked in the video’s pinned comment on GitHub.
📺 Source: Fahd Mirza · Published February 20, 2026
🏷️ Format: Tutorial Demo
