Redis Queues: The Glue Between Your AI Agent Tasks
How Redis lists, sets, and TTLs became the simplest coordination layer for my multi-step AI automation pipeline.
Notes on building AI-powered tools, automation, and shipping fast.
How Redis lists, sets, and TTLs became the simplest coordination layer for my multi-step AI automation pipeline.
How I host ten services on a single $10/month VPS using Docker Compose, Traefik, and Cloudflare — with automatic HTTPS and zero-downtime deploys.
How I built a single-command cross-posting pipeline that publishes to my blog, Dev.to, and Hashnode simultaneously — and why canonical URLs matter more than you think.
What happens when AI cuts the cost of a website redesign from 15 hours to 15 minutes? Spec work becomes a scalable sales strategy.
How a tracking pixel, a webhook, and an AI agent replaced my CRM — for nearly zero cost.
Two scheduling patterns for autonomous AI agents — when to use periodic heartbeat polls vs precise cron jobs, and how to avoid the overlap trap.
Every AI agent session starts from zero. Here's why flat markdown files beat databases and vector stores as a persistence layer for autonomous agents.
When your AI prospect finder starts returning mostly duplicates, that is not a bug — it is one of the most useful signals your system can produce.
How a single self-hosted Convex instance replaced six separate backends and became the unified API layer for everything I build.
Building a sub-200ms voice AI agent by bridging Twilio Media Streams to OpenAI Realtime API — no STT/TTS pipeline needed.
How I built a fully autonomous blog publishing pipeline — from cron-triggered AI writing to cross-posting on Dev.to and Hashnode, with no human approval step.
When your AI prospecting system exhausts a market, the saturation signal matters more than the output count. Here is what I learned running 260+ automated prospect discoveries in South Florida.
The flashy AI demos get attention. But cron jobs, markdown files, and human fallbacks are what actually keep autonomous systems running for weeks.
My AI prospecting bot found 260+ leads autonomously — then hit a wall. The lesson nobody tells you about automation: knowing when your system has done its job.
Every automated system has a natural ceiling. Here is how duplicate rate became the most important metric in my AI prospecting pipeline.
After two weeks of autonomous prospecting, my AI bot saturated South Florida and taught me that knowing when to stop is more valuable than scaling up.
When my AI prospecting system stopped finding new leads, it wasn't broken — it had saturated the market. Here's what that taught me about building automation that knows when to pivot.
I built an AI system that autonomously finds sales prospects. Two weeks and 260+ prospects later, it hit saturation — and that turned out to be the most useful signal of all.
Two weeks of autonomous AI prospecting: 256 leads, zero manual searches, and the lessons learned building a self-running lead discovery machine.
I let my AI agent autonomously build a 256-prospect pipeline over two weeks. Here is what I learned about deduplication, geography saturation, and the real value of AI-driven prospecting.
I write about AI, automation, and building things fast. No spam, just signal.
Get in Touch