The lab notebook
Field notes on AI engineering
824 articles · page 19 of 28
MCP vs Plugins: How AI Tool Integration Evolved
A practical breakdown of how AI tool integration moved from custom plugins to the Model Context Protocol, and what that shift means for engineers building with LLMs today.
Measuring Productivity Gains From AI Coding Agents
A practical, honest framework for measuring whether AI coding agents like Codex CLI actually make your team faster, without relying on made-up percentages.
Migrating from Manual QA to DeepEval: A Team's Journey
How one AI engineering team replaced spreadsheet-based prompt reviews with DeepEval's automated LLM test suite, and what actually broke along the way.
Building a Multi-Modal RAG System: Text, Images and Tables Together
A practical guide to building multi-modal RAG systems that retrieve and reason over text, images, and tables in a single unified pipeline.
Multi-Query RAG: Generating Multiple Retrieval Queries Per Question
Learn how multi-query RAG generates several retrieval queries per question to fix recall gaps that a single embedding search misses.
Multi-Server MCP Setups: Composing Tools from Several Servers
A practical guide to running multiple MCP servers in one client, handling name collisions, and designing agents that compose tools cleanly.
Multi-Tenant RAG: Isolating Data Per Customer at Scale
A practical guide to isolating customer data in multi-tenant RAG systems, covering metadata filtering, namespace strategies, and the failure modes that leak data between tenants.
Building an Onboarding Agent for New Employees
A hands-on guide to building an onboarding AI agent that answers new-hire questions, tracks tasks, and escalates gaps to humans.
Online vs Offline Evaluation: When to Test in Production
A practical guide to choosing between offline eval sets and online production testing for LLM applications, with code and decision criteria.
OpenAI Codex for API Integration Tasks
A practical guide to using OpenAI Codex CLI for real-world API integration work, from auth scaffolding to error handling and testing.
OpenAI Codex Approval Workflow: Reviewing Before You Merge
A practical guide to Codex's approval modes, sandbox policies, and review habits that keep AI-written code safe before it hits main.
OpenAI Codex for Bug Fixing: A Systematic Workflow
A step-by-step workflow for fixing bugs with OpenAI Codex CLI, from reproducing the issue to verifying the patch and preventing regressions.
OpenAI Codex in CI/CD: Automating Routine Code Tasks
Learn how to wire OpenAI Codex into GitHub Actions to automate reviews, changelogs, dependency fixes, and routine PR chores.
OpenAI Codex for Code Review Automation
Learn how to automate code review with OpenAI Codex CLI, from local diff checks to CI pipeline integration and custom review prompts.
OpenAI Codex Configuration: Project-Level Instructions
Learn how to configure OpenAI Codex CLI at the project level with AGENTS.md and config.toml so every teammate and agent run follows the same rules.
OpenAI Codex Cost Management and Usage Limits
A practical guide to controlling OpenAI Codex CLI costs, understanding usage limits, and avoiding surprise bills while shipping real work.
OpenAI Codex for Data Science Workflows
A practical guide to using OpenAI Codex CLI for pandas cleaning, EDA, feature engineering, and notebook debugging in real data science work.
OpenAI Codex for Documentation Generation
Learn how to use OpenAI Codex CLI to generate accurate, maintainable documentation directly from your codebase instead of writing it by hand.
OpenAI Codex for Frontend Prototyping
A practical guide to using OpenAI Codex CLI to scaffold, iterate on, and ship frontend prototypes faster without losing code quality.
OpenAI Codex for Legacy Code Modernization
Learn how OpenAI Codex CLI helps engineering teams modernize COBOL, Java, and legacy systems safely with AGENTS.md, ExecPlans, and parallel validation.
OpenAI Codex for Onboarding New Engineers
How engineering teams use OpenAI Codex CLI to cut new-hire ramp time by turning a messy codebase into an explainable, queryable system.
OpenAI Codex Prompting Tips for Better Results
Learn practical OpenAI Codex prompting tips — from four-part task framing to AGENTS.md — that turn vague requests into reliable, reviewable code.
OpenAI Codex for Refactoring Large Functions Safely
Learn how to use OpenAI Codex CLI to break down large, unwieldy functions into safe, tested, well-structured code without breaking production.
OpenAI Codex Sandbox Modes Explained: Auto, Manual and Full Access
A practical breakdown of Codex CLI's sandbox and approval modes, how they enforce filesystem and network limits, and when to use each one safely.
OpenAI Codex Security Considerations for Enterprise Use
A practical breakdown of Codex's sandboxing, approval modes, data handling, and the enterprise controls security teams should verify before rollout.
OpenAI Codex for Solo Developers: A Realistic Workflow
A grounded, no-hype workflow for using OpenAI Codex as a solo developer, from scoping tasks to reviewing diffs before they ship.
OpenAI Codex Team Adoption: Rolling Out AI Coding Agents Safely
A practical playbook for rolling out OpenAI Codex across an engineering team without breaking security, code quality, or trust.
OpenAI Codex Terminal Workflow: Tips for Power Users
Practical CLI tips for OpenAI Codex power users covering flags, sandboxing, approval modes, and terminal-native workflows that speed up daily coding.
OpenAI Codex for Test Generation
A practical guide to generating unit, integration, and edge-case tests with OpenAI Codex CLI, including prompts, workflows, and pitfalls to avoid.
OpenAI Codex Common Errors and Troubleshooting
A practical, error-by-error guide to fixing the most common OpenAI Codex CLI failures, from auth loops to sandbox denials to MCP timeouts.