Fake CVEs and What They Tell You About Your Security ProcessThe Problem With Panic-Driven Security Last month, a CVE identifier for a supposedly critical PHP vulnerability made the rounds on Twitter and security forums. Teams scrambled. Slack channels lit up.4 min read·icentric-ai-and-automation.hashnode.dev2d ago
Building Stateful AI Agents: What MCP Changes for DevelopersBuilding Stateful AI Agents: What MCP Changes for Developers If you've been building AI-powered tools lately, you've probably hit the same wall I have: context fragmentation. Your LLM can reason brill4 min read·icentric-ai-and-automation.hashnode.dev2d ago
Why MCP Changes the Game for AI-Driven Outreach InfrastructureThe Problem With Building Outreach Tooling Today If you've ever built or maintained an outreach stack—whether for sales, partnerships, or recruitment—you know the pattern. You start with a CRM. Then y4 min read·icentric-ai-and-automation.hashnode.devNov 2
Treating CVEs Like Pull Requests: Why Vulnerability Triage Needs Code Review DisciplineThe Trust Problem in Vulnerability Management We've built entire workflows around not trusting external input. We sanitise database queries, validate API payloads, and treat user-submitted data as hos4 min read·icentric-ai-and-automation.hashnode.devNov 2
Why I'm Moving Production AI Workloads Off Third-Party APIsWhy I'm Moving Production AI Workloads Off Third-Party APIs Six months ago, our team was comfortably spending £3k/month on OpenAI API calls. Last month, that figure hit £18k. The trajectory was clear:4 min read·icentric-ai-and-automation.hashnode.devOct 26
