The lab notebook
Field notes on AI engineering
824 articles · page 13 of 28
LLM-as-a-Judge Explained: Biases, Calibration and Best Practices
A technical guide to LLM-as-a-judge covering its seven common biases, pairwise vs pointwise scoring, rubric design, and calibration.
How to Build an Eval Pipeline That Blocks Bad Deploys in CI/CD
A practical guide to wiring LLM evals into GitHub Actions so bad prompt changes get blocked before they ship, not after.
LLM Evaluation 101: Metrics, Methods and Tools for 2026
A practitioner's guide to LLM evaluation covering metrics, golden datasets, LLM-as-a-judge, and CI/CD integration.
MCP vs Function Calling: What's the Real Difference in 2026?
Function calling wires a tool to one model's API. MCP standardizes discovery and invocation across every client. Here's when each earns its keep.
MCP vs REST APIs: Do You Still Need Both in 2026?
MCP does not replace REST APIs — here is the architecture for running both together, and how to decide which one to build first.
Multi-Agent Systems Explained: Orchestrator, Swarm and Debate Patterns
A technical breakdown of orchestrator-worker, swarm, and debate patterns for multi-agent systems, with failure modes and code.
n8n AI Agent Nodes Explained: Building Your First Automated Agent
A hands-on n8n AI agent tutorial covering triggers, LLM nodes, routing, tool calls, memory, and error handling.
n8n LangChain Integration: Connecting Your Workflows to LLMs
A practitioner's guide to n8n's LangChain-style nodes — models, chains, agents, memory, and vector stores — inside real automation workflows.
n8n vs Zapier vs Make: The 2026 AI Automation Comparison
A practitioner's breakdown of n8n vs Zapier vs Make for 2026, covering pricing models, self-hosting, and AI-agent automation.
Open Source LLMs vs Closed Models: The 2026 Buyer's Guide
A practitioner's framework for choosing between open source LLMs and closed models on cost, privacy, latency, and quality.
OpenAI Codex CLI Tutorial: Setup, Config and First Task
A practitioner's walkthrough of installing, configuring, and running your first real task with OpenAI Codex CLI.
OpenAI Codex vs Claude Code: Which Terminal AI Agent Wins in 2026?
A practitioner's side-by-side of OpenAI Codex and Claude Code across setup, planning, sandboxing, and real workflows.
Prompt Engineering vs Fine-Tuning: When to Use Which in 2026
A practical decision framework for choosing prompt engineering, RAG, or fine-tuning based on cost, latency, and data reality.
RAG vs Fine-Tuning: Which One Actually Fixes Your LLM's Knowledge Gaps
A practitioner's decision framework for RAG vs fine-tuning, covering costs, maintenance, and when each actually fixes knowledge gaps.
Ragas Metrics Explained: Faithfulness, Context Precision and Recall
A practitioner's breakdown of Ragas metrics explained — Faithfulness, Context Precision, Context Recall, and Answer Relevancy, with worked examples.
Synthetic Test Set Generation with Ragas: A Practical Walkthrough
How to generate, filter, and blend Ragas synthetic test sets so your RAG evaluation actually covers your knowledge base.
Ragas Tutorial: Evaluating RAG Pipelines in 30 Minutes
A hands-on Ragas tutorial that shows you how to score faithfulness, relevancy, and retrieval quality in your RAG pipeline in half an hour.
Ragas vs DeepEval: Choosing Your RAG Evaluation Framework
A startup's support bot forces a real choice between Ragas and DeepEval — here's how to pick based on your growth stage, not a feature checklist.
Self-Hosting n8n: A Step-by-Step Production Setup Guide
A practitioner's guide to running n8n in production: Docker Compose, Postgres, HTTPS, backups, and safe upgrades.
8 Types of RAG Architectures Explained: Naive to Agentic RAG
A practitioner's breakdown of the 8 types of RAG architectures, from naive vector search to agentic RAG, with tradeoffs for each.
What Is an AI Agent? A Practical Definition for Engineers in 2026
A clear, working definition of what an AI agent actually is — with code, failure modes, and when to skip building one.
What Is MCP (Model Context Protocol)? The Complete 2026 Guide
A practitioner's guide to MCP — what it is, why it exists, and how the client-host-server model connects AI agents to real tools and data.
What Is the Hermes Agent? Inside TeachYou's 30-Day Agent Bootcamp
A practical breakdown of the Hermes Agent, TeachYou's 30-day build-in-public bootcamp for production-grade AI agents.
Why Most AI Agent Projects Fail (And How to Avoid It)
Most AI agent projects don't fail because of the model — they fail because of process gaps that show up the moment a demo becomes a product.
Agent-to-Agent (A2A) and MCP: How They Fit Together
A2A lets agents talk to other agents; MCP lets an agent talk to tools and data — here's how the two protocols work together in production.
A/B Testing LLM Prompts in Production
A practical guide to A/B testing LLM prompts in production, from traffic splitting to statistical significance and rollback strategy.
What Makes a Good Agent Capstone Project?
A practical framework for scoping, building, and shipping an AI agent capstone that actually proves you can build in production.
Agent Cold Start Problems: Why First Requests Feel Slow
The first request to an AI agent is almost always the slowest one, and understanding why is the first step to fixing it.
Agent Cost Optimization: Cutting Token Spend Without Losing Quality
A practical guide to slashing LLM agent token spend through caching, model routing, and context pruning without hurting output quality.
Agent Fallback Strategies: What Happens When the LLM Is Unsure
A practical guide to designing fallback paths for AI agents so uncertainty becomes a signal to handle, not a failure mode to hide.