AI & Automation Hub
After cloud native comes the world without your own servers: workloads run serverless at the edge of the network, databases and queues are platform services, and AI tools take over routine work in pipelines and operations. This hub shows what that world already looks like in practice: reproducible implementations with Cloudflare Workers and Claude, always with an eye on security, cost, and operations rather than glossy demos.
Topics in this hub
Latest articles
Up to 10 Million Free Tokens per Day: Using OpenAI’s Data-Sharing Program with Cost Guardrails
OpenAI gives organizations that share their API traffic for training a daily free allowance: depending on the tier, up to 10 million tokens. With prepaid credit, project limits, and a token budget in the code, usage remains free indefinitely.
Claude Desktop keeps crashing: “GPU process gone” with exit code 101457950, cause and solution
The Claude Desktop app on Windows exits completely with “GPU process gone: exitCode 101457950” (0x060C201E), often followed by the Store app repair dialog. The complete causal chain: Code Integrity blocks vk_swiftshader.dll, Chromium’s fallback chain runs out, and the built-in self-termination exits the app. With a permanent solution (switching to the classic installation without MSIX), self-diagnosis via the event log, and analysis down to the minidump.
Run your own newsletter with Cloudflare Workers and D1
The open template provides signup, unsubscribe, queue, and database in your own Cloudflare account. A deploy button sets up Worker, D1, and CI without a local server.
Running Claude Code Securely on Your Own VPS
A hardened Debian VPS keeps Claude Code sessions persistently available. This guide covers everything from user accounts and SSH keys to firewalls, data hygiene, tmux, and secure access from an iPhone.
