Haunted Browsers, AI Cinema: Directing Your Story with Gemini CLI en

Since it came out, Gemini CLI has revolutionised the Open Source AI world.

2 demos:

1. The "Haunted Browser": a web developer's dream. See how MCP Playwright, automates tedious web tasks like clicking buttons, logging in, and navigating sites - a "curl v2".

2. "AI Cinema": a creator's studio. Watch as we use the CLI to direct and generate entire, multi-step video storylines from a single prompt. 🤖 Two powerful demos, one tool, and we'll show how both can be automated in the cloud with GitHub Actions.

Beyond Blame: The Art of the Postmortem en

Let's be honest: postmortems are hard. Too often, they become a political exercise in "Cover Your Assets," failing to uncover the real root causes. How do we move from a culture of blame to one of genuine learning?

As an Italian SRE expert at Google—the company that pioneered this space—I've spent 6 years evangelizing SRE principles across Europe.

In this session, we will dissect the anatomy of a truly effective postmortem. Forget dry theory; this is about learning from the front lines. Through real-world stories and hard-won insights, you will learn when a postmortem is essential, how to adapt Google's foundational techniques for your team, and how to transform failures into powerful business insights.

This talk is for every operator and leader dedicated to building a culture of real resilience. Let's make your next outage your most valuable lesson.

SRE Gone Wrong: Top 7 Errors (and How to Avoid Them) en

Think SRE adoption is a smooth journey?

Think again!

In this eye-opening talk, Riccardo reveals the top 7 blunders companies make when implementing SRE, drawing from years of hands-on experience and the insights captured in the "Enterprise Roadmap to SRE" book.

Discover the hilarious (and sometimes painful) mistakes that can derail your SRE journey, and learn how to sidestep them with practical, actionable strategies. Key takeaways:

* SRE Essentials: Demystify SRE and understand its core principles.
* The Top 7 Errors: Learn from the most frequent and impactful mistakes companies make when adopting SRE.
* Unlock Success: Leave equipped with the knowledge and strategies to navigate the SRE adoption minefield and achieve true reliability and efficiency.

Pragmatic AppMod: take an old PHP project and make it 12-factor compliant on GCP en

Do you have a pet project from 10-20 years ago, and you barely remember what it does? When deploying to production was simply overwriting some files over FTP?
How about embarking on a journey to App Modernization to leverage GCP powerful tools to make it 12factor-compliant? Riccardo will walk you through a pragmatic trip through the modernization of an old app:
* Moving to a git repo
* 📦 Dockerize the app
* Import DB to Cloud SQL
* Extracting vars (and DB info) into a .env and leverage Secret Manager to manage secrets.
* Build/Release through Cloud Build [and Cloud Deploy]
* Secure the DB removing public IP (via private service)
* Deploy to Cloud Run 🏃
* Observe logs / create custom metrics
Also introduce some common pitfalls and how to solve them.

Incident Management at Google en

SRE practices extend to more than just observability. They also include: Capacity Planning, Change Management, being prepared for disasters.

In this session Riccardo will walk us through "IMAG", or how Google does Incident Management.

Goodbye, deployment headaches: Cloud Deploy and Vertex AI unite en

Continuous Deployment can be a roadblock in the MLOps lifecycle, often requiring custom pipelines and complex configurations. Solution? The new integrations of Cloud Deploy and Vertex AI revolutionizes machine learning (ML) deployment by automating the entire process, and makes it easy to roll back through idempotent releases. The groundbreaking integration of Cloud Deploy and Vertex AI lets you test, validate, and deploy your ML models in minutes, without writing a single line of code.

Agents on a Plane: A Deep Dive into Building a Real-Time Travel Agent ✈️ en

In 2025, an LLM agent needs to do more than just talk. It needs to do. This presentation is a hands-on look at building a truly useful AI travel agent with a multi-agent framework.
Forget abstract concepts; we're diving straight into the code of a live demo that queries real APIs for flights, hotels 🏨, and AirB&Bs. We will cover how to structure a multi-agent system to handle complex, real-world tasks: validating budgets, managing user preferences, and visualizing results. You will learn how to implement an advanced, modular system that can be deployed on any cloud with any LLM.
This is a technical L200-L300 session designed for developers who want to see how to coordinate agents and tools to create a seamless, end-to-end user experience.

[Talk+Demo]

News Crawler via Langchain.RB and Gemini APIs en

How can we get an LLM to be updated to today’s news?
Gen AI is great at answering questions.. from the past. After the LLM was trained, all you can do is RAG.
How about crawling the web for latest news with Gemini for multimodal extraction and offering summarization by your favorite topic?
It all gets more exciting thanks to Andrei’s langchainrb gem.

GenAI: Kids Stories That Come to Life en

Ever wondered if GenAI can create customised kids stories and also illustrate them? How about translating them to your favourite language, and read them loud for you in Spanish, Japanese, even German? All of this is possible through Google Cloud APIs.

In this blast from the past, Riccardo unites latest Google GenAI technologies with a powerful framework (Ruby on Rails) to demonstrate

Agenti in Volo: Un'Analisi Approfondita sulla Creazione di un Agente di Viaggio in Tempo Reale ✈️ en

Nel 2025, un agente basato su LLM deve fare più che parlare. Deve agire. Questa presentazione è uno sguardo pratico alla costruzione di un agente di viaggio IA realmente utile, utilizzando un framework multi-agente.

Dimentica i concetti astratti: ci tufferemo direttamente nel codice di una demo live che interroga API reali per voli, hotel 🏨 e AirB&B. Vedremo come strutturare un sistema multi-agente per gestire compiti complessi e reali: validare budget, gestire le preferenze dell'utente e visualizzare i risultati. Imparerai come implementare un sistema avanzato e modulare che può essere distribuito su qualsiasi cloud e con qualsiasi LLM.

Questa è una sessione tecnica di livello L200-L300, pensata per sviluppatori che vogliono scoprire come coordinare agenti e strumenti per creare un'esperienza utente fluida e completa.

Gemini CLI: emerging agents en

Since it came out, Gemini CLI has revolutioned the Open Source world with its 50K stars in few weeks.
But how to have emerging Agentic behavior from a CLI command running in localhost? Is this even possible?!?

Riccardo will guide us through an exciting demo showing a number of agents lurking in local FileSystems doing things and keeping it local.

What happens in the file system... stays in the file system.

This will be a "WOW" effect demo where multiple agents are spun up.
High speed internet required.

[WS] Let's vibe code together an app with Gemini CLI and MCP en

You probably have heard about Gemini CLI https://github.com/google-gemini/gemini-cli .

But have you tried all of its powerful features, like GEMINI.md, custom commands, extensions, MCP, GEMINI.md?

In this session, all you need is a computer on the internet and some creativity!

Riccardo will guide you through the rest. from `npm install` to a full fledged and well documented vibe-coded application. After the session, you'll have a fun app, but also you'll have learned a number of secret tricks to leverage your Gemini CLI.

All you need is a computer!

This is a 2-4h workshop, hopefully a 4h one.

Stop Grepping, Start Reasoning: Skill-Based Agentic SRE 🐉 en

Let’s be honest: debugging production at scale is soul-crushing manual labor.
At Google, we decided making thousands of SREs act like human search engines in 2026 was a bug, not a feature. Enter the Agentic SRE Extension—a skill-based framework that lives in your terminal and actually knows its way around Kubernetes.
🚀 In this high-energy deep dive, I’m showing you exactly how we use modular skills and MCP to chain diagnostic tools, analyze metric regressions, and execute safe mitigations (Rollbacks, Throttling) without the "deleted production" anxiety.
🔥 We’ll dissect the "Outage Investigator" agent's logic loop, see it draft a technical postmortem in seconds, and discuss why we built this as a portable framework whose skills can be easily leveraged by any modern AI harness. You’ll leave with the code to wire up your own Kubernetes stack and let the agent's skills do the heavy lifting while you drink espresso.
No fluff, no 101s. Just AI agents, specialized skills, and less toil.

Target Audience: SREs, DevOps Practitioners, Platform Engineers, and Developers interested in AI-assisted operations.

Preferred Session Duration: 30 minutes (40m if I can do a demo).
Technical Requirements: Internet connection to execute a live terminal-based outage investigation demo on a cloud Kubernetes cluster.
First Public Delivery: This will be the first public deep-dive into the skill-based agentic framework methodology and live demonstration of the "Outage Investigator" logic loop.

Did you know I could write in Rust? Neither did I! en it

Think you need to be a Rust wizard to write powerful command-line interfaces? Think again!

This session is a fun, emojiful journey into how I built CLIs in Rust without actually knowing the language.

We’ll explore the unexpected joy of learning the build and test suites of a language I knew nothing about and the art of babysitting code with AI.
Whether you're a software engineer, an automation enthusiast, or a home server dreamer, join me to see how AI made languages a commodity and the lessons I learnt along the way. 🥑

This is a screenshot-intensive proof that writing a good CLI has nothing to do with knowing a language and more about having the right specs to make the CLI user and agent friendly.

Sapevate che so programmare in Rust? Tranquilli, non lo sapevo manco io! 🦀 en it

Pensate che serva essere uno Stregone Supremo di Rust (con tanto di barba bianca e mantello) per scrivere potenti interfacce a riga di comando? Ricredetevi!

Questa sessione sarà un viaggio tragicomico, abusivamente pieno di emoji 🚀🦀, in cui vi mostrerò come ho tirato su delle CLI in Rust senza avere la più pallida idea di come diavolo funzioni il linguaggio. (Spoiler: di solito il mio habitat naturale è scrivere spaghetti code in Ruby, quindi potete immaginarvi lo shock termico).

Esploreremo l'inaspettata "gioia" — leggi: disperazione mista a sudore freddo — di scontrarsi con le build e le test suite di un ecosistema a me totalmente alieno, e la nobile, antica arte di fare da babysitter al codice generato dall'Intelligenza Artificiale (sperando che non dia fuoco al computer).

Che siate ingegneri del software navigati, fanatici dell'automazione o sognatori che coccolano il proprio home server in cantina, unitevi a me! Scopriremo insieme come l'IA abbia ormai reso i linguaggi di programmazione un bene di consumo (intercambiabili quasi quanto i calzini) e vi racconterò tutte le assurde lezioni che ho imparato sbattendoci la testa. 🥑

Questa è la prova schiacciante — e pericolosamente zeppa di screenshot — che per sfornare un'ottima CLI non serve avere la benché minima idea del linguaggio di programmazione che state usando. Il vero segreto? Avere le specifiche giuste per rendere la vostra creatura a prova di utente (e di agente IA), senza che nessuno si faccia male nel processo!

Build Your First AI Cloud App with Rails 8 & Gemini en

You don't need to know Ruby to join this workshop — just bring your laptop and curiosity.

In 90 minutes, you'll go from zero to a working AI-powered chat application that generates real-time responses and AI artwork, all running on your machine. No slides, no theory dumps — just code, experiments, and that "wow, I built this!" feeling.

We'll use **Rails 8** (the latest version, with built-in background jobs and real-time updates out of the box) and **Google Gemini** to build a chat app that talks back intelligently and creates images on the fly. Then we'll turn it into an **MCP server** — the new open standard that lets AI assistants like Gemini CLI interact directly with your app's data using natural language.

**What you'll walk away with:**

- A working AI chat app on your laptop with real-time streaming responses
- Hands-on experience with **MCP (Model Context Protocol)** — the hottest interoperability standard in AI tooling
- Your first custom MCP tool: teach an AI to query your database in plain English
- A clear path to deploy everything on **Google Cloud Run** with zero DevOps headaches

**What you need:**

- A laptop with Docker installed
- A free Google AI Studio API key (we'll help you get one!)
- Zero Ruby experience required — we'll guide you step by step