Prodigal reposted this
At Prodigal, we’ve always been an AI-first company, embracing new tools and AI agents to boost our productivity. It's common to see our Devin working side by side with our engineers. And the ubiquitousness of AI tools like Cursor and Claude Code as part of our daily workflows 🤖 Personally, I've always been fascinated by LLMs, and have always wanted to find a way to make my DevOps workflows more conversational. In the past, my attempts involved a lot of API calls, trial-and-error testing, and a messy patchwork of solutions. Eventually, the complexity led me to give up and revert to the old manual way. That was until my friend Dhruv Grover 🙌 started talking about MCP Servers and how they were the next big thing in GenAI. After hearing his talk and seeing his experiments, I realized this could be the breakthrough I needed to finally make my workflows truly conversational. The early results from my experiments with Amazon Web Services (AWS)’s MCP servers are looking promising. Here's a quick snapshot: with just a single prompt in Claude Desktop, I was able to analyze our data transfer costs over the past 10 days and identify opportunities to save 30-50% in costs. This is just the start! We’re also exploring other ways to leverage MCP servers and LLMs to: ⏰ Analyze incidents based on CloudWatch alarms, metrics, and logs 💰 Suggest and apply cost-saving recommendations (like EBS right-sizing, lifecycle policies on S3, etc.) ⚙️ Manage and change infrastructure using natural language, while maintaining IaC in source control 🛡️ Boost security and compliance, ensuring our apps meet industry standards We’re already seeing great results, and I truly believe this opens up huge potential for automating DevOps/MLOps, all while optimizing for both efficiency and security. I’ll keep posting updates as we continue experimenting and discovering new breakthroughs. Would love to hear what others are doing with MCP! #MCPServers #AWS #DevOps