Welcome to The Agent Lair

Where AI agents come to life. Explore the latest in autonomous systems, intelligent automation, and the future of AI-driven solutions.

AI AgentsAutonomous SystemsLLM EngineeringAgent Architecture

Featured Post

FeaturedArize

How Arize helped us run 45 experiments to build the most accurate food logging in the market

We thought our AI food logging was accurate. Arize showed us it was 52%. Here's how we used Arize's dataset, experiment, and tracing features to run 45 controlled tests — and what the data showed at each step.

1 day ago

22 min read

Read article

Recent Posts

GitAI

2 months ago

5 min

How git blame powers Sticky Note's attribution engine

Sticky Note uses git blame, Git Notes, and a three-tier fallback pipeline to figure out which AI session wrote which lines of code. Here's how the whole thing works.

Engineering

Copilot CLIClaude Code

2 months ago

19 min

Making Sticky Note work with Copilot CLI

We built Sticky Note for Claude Code first. Getting it to work in Copilot CLI meant dealing with a bunch of runtime differences nobody warns you about.

Sticky Note & branch switching: design notes

Sticky Note stores session data in git-tracked files that change on every AI tool call. Here's how we designed a three-layer defense to make branch switching work without losing your data.

A/B testing system prompts on Arize: how we track MOBA's chatbot quality

How we use Arize AX to run controlled prompt experiments on MOBA's nutrition chatbot — deterministic variant assignment, span tagging, and eval score comparisons across system prompt versions.

Production evals on Arize: what the docs don't tell you

We spent a day getting LLM-as-judge evaluators running on production MOBA traces. Zero successes → 900 successes. Here's every mistake we made and what finally worked.

Stop Guessing If You're Recovering: The Science Behind MOBA's Sleep Analysis

I built this because I couldn't sleep for two months—and realized every athlete is flying blind on the most important part of training. Here's the science behind every number MOBA shows you.

Product

n8nworkflow automation

3 months ago

11 min

I Got Tired of n8n Workflows That Silently Broke

Most workflow tools stop at 'deployed successfully.' I kept finding empty Google Sheets at 11pm because nodes failed without telling anyone. So I built a skill set that builds n8n workflows, tests every node's output, and debugs them when data disappears.

Smart Calorie Targets: How We're Revolutionizing Nutrition for Athletes

Finally, calorie targets that actually understand your training. How we built a system that learns your sport, detects your training phase, and predicts your workouts.

Deploy ClawdBot to Azure Container Apps: Your 24/7 AI Assistant in 30 Minutes

Deploy ClawdBot - the open-source personal AI assistant - on Azure Container Apps with a single command. Get built-in security features, automatic TLS, and 24/7 operation for ~$40-60/month.

Building an Autonomous AI Agent Loop with Ralph and GitHub Copilot CLI

Learn how to use the Ralph pattern with GitHub Copilot CLI to build entire projects autonomously. We built Rekoma-Context—19 user stories, 109 tests—with zero manual intervention.

AI Engineering

React NativeTypeScript

5 months ago

9 min

Shipping MOBA: Cleaning Up All the Vibe Coding

How we removed 6,400+ lines of dead code from our React Native app. A practical guide to cleaning up the chaos of rapid prototyping.

Debugging Private Networks When Public Queries Are Blocked: Azure SRE Agent + MCP

How Azure SRE Agent investigates resources in isolated VNets using a VNet-integrated MCP server—when Log Analytics queries are blocked by Private Link

Technology