Using an AI agent to navigate an undocumented Kubernetes repo
A concrete example of using an AI agent to explore a GitOps repo, find a production-only CronJob, map confusing stage/staging overlays, and generate infrastructure documentation.
Enrico Teotti | Technical & Product Consultant
Technical coach and product management consultant with 20+ years in software development across the US, Australia, and Europe. I help teams through strategy, facilitation, and hands-on guidance.
Latest posts
Precision, Recall, and F1: A Simple Explanation
Precision and recall separate two kinds of classifier error that a single accuracy number hides. Here's what they mean, when each one matters, and how F1 combines them.
How to build evals for LLM prompts in production
A practical guide to building eval systems for LLM prompts in production: project structure, test datasets, output formats, and the blind spots most teams miss.
Zombie Barbies: AI writes prettier tickets than your PM
Vague tickets stall software development. Polished tickets without strategy are Zombie Barbies. Five templates for features, bugs, spikes, chores, and non-user work — plus an LLM prompt to generate them via MCP.
80 prompts, zero evals: fixing an LLM pipeline after launch
A client shipped an 80-prompt LLM pipeline with no evals, no audit trail, and no dashboards. By month five, operators were throwing away 80% of the output. Here's how I helped the team fix it — dashboards, deterministic safeguards, prompt isolation — while operators were still using the system.
One Year with AI in Production: The Autonomy Myth and What Actually Works
AI agents won't ship features while you sleep (yet), but they will change who can contribute to your codebase
From 68% Recall to 95%: Fixing a Job Classifier Without Increasing LLM Spend
A legacy LLM classifier was missing 32% of the jobs it was built to find. Here's how I replaced it with an ensemble of heuristics, ML, and a multi-step LLM chain — without increasing API costs.
LLM with MCPs assessing application performance
How connecting an LLM agent to Sentry, Posthog, and a production database helped me cut a 32-second p95 to 600ms
How to Write Bug Reports That Engineers Can Work On
A practical guide to writing bug reports engineers want to pick up — with templates, examples, and prioritization guidelines.
Faster bug triaging with AI LLMs via MCP
AI LLM with your production database and application source code can help you debug and triage issues in minutes
Generate product insights with AI Agents, DB MCPs and your app sourcecode
Connecting AI agents to your database via MCP? You're missing half the picture. Add your app's source code to unlock domain-aware product insights.
Shopping list to North Star
Your product roadmap is a shopping list. Here's a workshop to turn it into a north star with leading indicators you can actually steer by.
LLM and MCP: A primer
An LLM is a brain thinking with its eyes closed. MCP lets it see and act. A simple guide to the building blocks of modern AI agents.
8 Standards for Shipping Production LLM Features
Shipping LLM features without standards means flying blind. A practical checklist: test datasets, evals, budget limits, traces, and version control.
Is it virtual or online facilitation?
People tend to use virtual and online interchangeably but they carry different meaning.
Key questions to ask in a Sprint Retrospective
What are the key questions that you ask in your sprint retrospectives?
Star Wars Agile Retrospectives a Principle 12 Story
Can a metaphor help your Agile Retrospective? Here's my take with Star Wars. Do we want this Agile Retrospective outcome to destroy a tie-fighter or defeat the empire?
Cynefin helps Agile Retrospectives
Do you feel like your Agile Retrospective is only tackling low hanging fruit and missing out on real improvements? Have you ever wondered if in your retro you're dealing with clear, complicated or complex issues? In this post I write how Cynefin can help with that.
Questions to ponder when Agile Retrospectives are missing reflection
In this post I share the chat I had with folks from HSD sharing my wicked issue of Agile Retrospectives missing out on reflection
How I ask for questions while facilitating online
How I ask for questions and antipatterns
From software engineer to visualizer
In this post I describe my journey in to the visualization world with links to resources that helped in my progression and that I wish I had on day 1
The law of two feet in open space online
This article is for facilitators holding space and organizers of Open Space events that moved or want to move online.
Video call from your LAN during COVID-19
What happens before an agile retrospective
Big picture event storming. My notes from the masterclass.
This is a summary of what I learned--circa 15 months ago--about big picture event storming during a two day workshop with Alberto Brandolini creator of Event Storming.
Event storming one year later
This is a summary of what I learned--circa 15 months ago--about big picture event storming during a two day workshop with Alberto Brandolini creator of Event Storming.
Make your retrospective facilitation even more effective? Here's 4 things to look at
Many factors--specific to your context--can reduce the effectiveness of a retrospective and cause frustration and mistrust in the group. In this brief blog post I will describe how psychological safety, purpose, preparation and collaboration can help make your retrospectives even more effective**.
Remote visual facilitation with iPad pro
Brief explanation of how I do remote visual facilitation with iPad pro and screen sharing.
Challenge Kerth's prime directive when you set the stage of your retro
This post assumes you know what a retrospective is and focuses on an activity to set the stage challenging what Kert's prime directive means to the group.
Analyze Kerth prime directive
This post assumes you know what a retrospective is and focuses on Kerth's prime directive to ensure the facilitator can answer common challenges and confidently explain it to the group.
Measure success of your agile retrospective
Retrospectives should be a core practice for any team but you must have a way to measure their success and team engagement otherwise they can become just another meeting. In this article I'll explain why they're both critical to retrospective success.