Agentic AI in DevOps, without the hype
Weekly perspectives from engineers building and running AI on real infrastructure.
#13 — cmux vs iTerm with Viktor Vedmich
cmux workflows for faster Claude Code sessions, Kiro tradeoffs, and AI token cost reality
cmux terminal workflows: when iTerm slows down AI-agent work, Viktor Vedmich shows how workspaces, session restore, and socket APIs make Claude Code more practical. Andrey Devyatkin and Fernando Gonçalves talk with Viktor Vedmich about his terminal-first agentic stack, Obsidian vault setup, Claude Code on Bedrock, AWS Kiro, spec-driven workflows, MCP search tools, rising token bills, and why open-weight model competition matters for DevOps teams.
Read
#12 — Semantic Layers, Context Layers, and Agents That Stop Guessing
Why agents need live system context and shared meaning before they can act with confidence.
13 min read
#11 — Base of Record for Intelligent Systems
Why AI agents need a second brain before they can safely understand engineering systems.
10 min read
#10 — What Changed in Our Daily AI Workflow
Three engineers compare the habits, guardrails, and timing shifts reshaping everyday AI work
9 min read
#9 — Code with Claude: Routines, Agents, and the AWS Catch
Anthropic’s coding push gets real, but its new AWS route hides a compliance trap.
9 min read
#8 — DevOps Jobs Agentic AI Can Actually Do
Where agentic AI helps DevOps today, and where state, cost, and production risk still block it.
8 min read
#7 — Agent Memory: When It Helps, When It Hurts, and How to Manage It
Persistent memory can make AI agents smarter — or poison every session. Here's how the memory lifecycle works and when to skip it entirely.
8 min read
#6 — The Big AI Squeeze
As AI discounts fade, teams face the real economics of subscriptions, local models, and sustainable automation.
12 min read
#5 — Stop Your Agent Before It Breaks Prod
How hooks turn unpredictable coding agents into controlled, auditable workflows
11 min read
#4 — Harness Engineering: What Claude Code Accidentally Taught Everyone
Why the tool layer, not the model alone, decides whether coding agents deliver or derail
7 min read
From Frustration to Product: The Story of B.O.R.I.S
How a broken experience with AI CLI tools led us to build a context layer that actually knows your infrastructure.
5 min read
#3 — Skills, Powers, SOPs
Agent skills sound like magic until someone uploads malware to the public hub.
7 min read
#2 — The Tool Layer: What Makes Agentic AI Possible
Context windows, MCP overhead, and why micromanaging your AI agent makes it worse.
7 min read
#1 — AI in DevOps, 2022 to 2026: From Autocomplete to Action
How AI went from clever autocomplete to agents that can act on your infrastructure — and why context is the missing piece.
7 min read