25 au 27 février, 2026
Montréal, Canada
Conférence Intelligence Artificielle
La liste des présentations est sujette à changement sans préavis.
Intelligence Artificielle
Leverage edge AI to deliver instant, private, and offline-ready intelligence in your mobile apps. We’ll explore practical ways to use on-device foundation models like Gemini-nano and Apple Intelligence and explain the balance between the powerful benefits of edge AI and its unique technical trade-offs to ship smarter and delightful mobile experiences.
Intelligence Artificielle
Migrating legacy .NET apps from Windows to Linux can be complex and costly. This session shows how generative AI can simplify modernization by automating dependency analysis, version upgrades, and code transformations cutting time and expense. Learn how to overcome scaling challenges, reduce licensing costs, improve security, and free your teams to focus on delivering business value instead of manual migration tasks.
Intelligence Artificielle
AI agents and automated workloads are now interacting directly with production systems, making API calls, triggering automation, and even writing data at machine speed. This creates new risks: over-privileged tokens, opaque decision-making, and lack of real-time control. In this session, we show how to secure these workloads using identity-aware proxies, OAuth 2.1, and Zero Trust policies.
Intelligence Artificielle
Coding Agents succeed when you give them structure. This session runs like a workshop: we’ll demonstrate how to build a feature with coding agents by defining a spec, managing context, and breaking the work into smaller subtasks. Along the way, you’ll see real prompts, failure cases, and handoff rhythms that keep quality high while boosting delivery speed.
Intelligence Artificielle
This presentation examines how AI automation through MCP servers revolutionizes development team management and project workflows. Attendees explore practical implementations of AI-powered ticket assessment, automated bug resolution, UI design, and Full SDLC management. Participants gain actionable strategies to implement intelligent automation that reduces overhead while improving productivity and delivery speed.
Intelligence Artificielle
LLM as a judge uses a language model to evaluate responses based on criteria you define, and prompt optimization is the process of refining the evaluation prompt to ensure the LLM judge is consistent, accurate, and fair in its scoring. Tools and techniques exist to automate this optimization, helping to capture complex evaluation criteria even with limited human supervision, improving performance, and reducing costly manual iterations.
Intelligence Artificielle
As AI becomes essential for enterprise apps, Java developers need to add intelligent features without rewriting their stack. This introduction to Spring AI shows how to bring generative AI to Java applications through practical examples. Learn to build chatbots, implement RAG for enhanced context, and use MCP for AI orchestration. Write model agnostic code that works with cloud and local LLMs. See how Spring AI's familiar abstractions make adding
Intelligence Artificielle
In this presentation I will be showing how I use PHP, Go, and Home Assistant to automate many tasks in my office and house.
We'll be looking at how to create AI (and classical) models for Home Assistant, with MQTT from both PHP and Go. With this we can create complex scenarios for complex situations, such as trying to figure out how dark and cold/warm it is going to be in a room.
We'll be looking at how to create AI (and classical) models for Home Assistant, with MQTT from both PHP and Go. With this we can create complex scenarios for complex situations, such as trying to figure out how dark and cold/warm it is going to be in a room.
Intelligence Artificielle
AI’s potential in software development goes far beyond code generation or IDE chatbots. This talk explores five AI capability families (Learning, Coding Assistants, Automation & Agents, Planning, and Data Exploration) that can reshape the CI/CD lifecycle. From process assessment to VM migrations, centering AI integration around developers throughout the entire lifecycle of planning, coding, deployment, and monitoring unlocks greater ROI.
Intelligence Artificielle
Explore how Gemini 3 Pro powers multimodal generation (text, image, audio, video) and real-time agents, plus effortless “vibe coding” in Google AI Studio. Dive into the Gemini Live API for real-time bidirectional voice and video conversations with tone-aware replies, tool use, RAG, and session memory. Together we’ll cover prompt-to-code flows, media generation, voice-first use cases, and the next generation of MCP-driven agentic AI using Angular.
Intelligence Artificielle
This hands-on session will guide developers on how to build agentic AI systems for real-time business research using fragmented, fast-changing information. Using Reka Research API with C# to build AI assistants that search across multi-source and generate grounded, traceable answers.
Intelligence Artificielle
In this session, you will learn how to build voice agents. We will examine the elements and patterns of building voice agents, including full and half-cascade models, native speech models, and others. We will see how to run them locally and on the cloud. We will also explore common challenges in terms of getting reliable transcripts and tool calling, deployment infrastructure and considerations of how to deploy and scale voice agents.
Intelligence Artificielle
In the past year, we’ve seen a fundamental change: developers and enterprises are moving away from proprietary, closed-source models. To save costs, prioritize privacy, and allow for customization, they are building, testing, and deploying their own open models. Join us as we demo the tools for model serving/scaling, adding unique data, agentic workflows, evaluation/optimization, and more- built on Linux and Kubernetes!
Intelligence Artificielle
This session is a practical tour to compare different SDKs (OpenAI, Microsoft AI Extension, etc.), ergonomics, setup when trying to use AI from a .NET application. Let's talk about what worked, what didn’t, and lessons learned from the little gotchas you hit along the way.
Intelligence Artificielle
We’ll embed a copilot into a Next.js app that actually reads UI context (form values, selections, route state) and takes approved actions. We’ll connect chat to real capabilities via typed tool calls, gate risk with consent prompts. You’ll see simple routing between chat, tools, and retrieval—and how to fail safely. Attend this presentation, walk out with a understanding context?decide?act?audit template you can drop into your product.
Intelligence Artificielle
Cette session est une plongée technique dans la création d'un pipeline de diagnostic assisté par IA. Nous verrons comment enchaîner trois modèles de vision par ordinateur pour analyser des radiographies et détecter pneumonie, COVID-19 et tuberculose. Puis, nous utiliserons les résultats pour interroger un LLM via une architecture RAG afin de générer un rapport préliminaire. Un cas d'usage de bout en bout qui montre la puissance de la combinaison.
Intelligence Artificielle
AI isn’t coming for the future of software delivery, it’s already here, rewriting the rules in real time. Models are being trained, deployed, and scaled faster than ever, but while DevOps gave us speed and reliability for traditional applications, it wasn’t built for the messy, unpredictable world of machine learning. That’s where MLOps comes in, the next evolution of DevOps, engineered for the age of AI.
Intelligence Artificielle
Claude ici, GPT par-ici, Qwen par-là, mais comment les évaluer ? Cette session explore ce que mesurent et ratent des benchmarks comme HumanEval, MBPP, et DS-1000. Nous comparerons les métriques textuelles et basées sur l’exécution, verrons comment les benchmarks peuvent être biaisés, et expliquerons les jeux d’entraînement, de test et tests cachés. Vous apprendrez à juger la qualité et l’utilité d’un dataset, et même comment en construire un.
Intelligence Artificielle
Black-box models fall short when decisions demand clarity and trust. This session shows how Explainable AI make troubleshooting easier and reveals hidden patterns in data. From making financial risk models more transparent to helping retailers see what drives sales, we’ll explore real-world examples and focus on hands-on skills for building understandable models, including the use of Shapley values.
Intelligence Artificielle
On-device AI cuts latency, lowers cost, and keeps data private by running models locally. This talk shows how to ship real on-device AI that runs entirely on laptops and edge boxes. We’ll start with model selection. Next, we’ll add an OpenAI-compatible SDK and a simple UI.
Intelligence Artificielle
Being a developer these days means you’re not only writing code, but also a Linux admin, a DevOps engineer, you name it! Thankfully, the Model Context Protocol (MCP) has made it significantly easier for me to ask my natural language questions, such as “What’s consuming the most memory in our dev cluster right now?”, and quickly receive a response. Hint: It was Prometheus. Learn the most popular MCP Servers, and how to chain them together too!
Intelligence Artificielle
A few months ago, I joined a two-week online hackathon to push the limits of vibe coding and fell into browser-based 3D multiplayer. What began as a food fight game quickly escalated: tuning netcode, cramming shaders into Three.js, and helping build a makeshift metaverse of tiny, weird games stitched together like a 2003 webring. It spiraled into something strangely meaningful, like the early days of something much bigger.
Intelligence Artificielle
Are you looking to better understand how to better leverage claude.ai and why it behaves this way? Dealing with claude.ai is counter-intuitive for many people. I will present why claude.ai is behaving this way and how to use the ML perspective to better constrain him to do what you really want with prompts.
Intelligence Artificielle
What do they have in common? This session explores AI through the lens of neuroscience, drawing parallels between how the brain processes information and AI developments. We’ll highlight key definitions, milestones, and evaluation gaps in both fields, and consider how close we are to achieving AGI.
Intelligence Artificielle
Building supportive AI assistants requires the engineering of conversational context without dilution. We will cover the stages of conversation: setting initial context foundations; using summarizers and memory types (short/long/persistent) as conversations grow; and leveraging RAG for permanent knowledge and graph structures for dynamic information on entities. Includes practical implementation details like audio capabilities and observability.
Intelligence Artificielle
Demonstrates WARP 2.0's Model Context Protocol servers and Rules engine for complete development lifecycle automation. Attendees will master using WARP AI to build multi-tiered applications, configure Docker environments, and implement comprehensive testing frameworks through intelligent automation rather than manual coding. Participants gain practical skills to streamline their entire development process from initial concept through production.
Intelligence Artificielle
Large Language Models are everywhere, but how they actually work “under the hood” often remains a mystery. In this live-coding talk, we’ll skip the slides and build an LLM step by step, not just a toy model, but a state-of-the-art one. Along the way we’ll dive into tokenization, transformer architecture, small-scale training, and inference, showing how these components come together to power today’s most advanced models.
Intelligence Artificielle
In this talk, you’ll learn how to craft high-quality video content using Google AI. You’ll move from concept and prompt iteration, through image and layout design, to animated video generation and voice synthesis via Gemini, Nano Banana Pro, and Veo.
Along the way, you’ll see how to layer in intelligence, like tool integration, retrieval-augmented generation (RAG), and API calls, to make your short reels ready for publishing using Angular.
Along the way, you’ll see how to layer in intelligence, like tool integration, retrieval-augmented generation (RAG), and API calls, to make your short reels ready for publishing using Angular.
Intelligence Artificielle
LLMs are often pitched as coding co-pilots, but their true long-term value for engineering teams can be in tackling the tedious work nobody wants to do.
This session will present a selection of practical, "boring" LLM use cases, such as the enforcement of security policies and training. We'll have a critical look at what to trust and how to verify an LLM's output when it comes to code and knowledge transfer.
This session will present a selection of practical, "boring" LLM use cases, such as the enforcement of security policies and training. We'll have a critical look at what to trust and how to verify an LLM's output when it comes to code and knowledge transfer.
Intelligence Artificielle
AI tools are rapidly embedding themselves into every layer of the modern development stack, whether via design platforms, IDEs, chatbots, review services, MCP servers, or CLI tools. But what happens to your sensitive data and code when these tools are involved? What are they really up to behind the scenes, and what risks do they introduce?
We'll take a practical tour of threats, vectors, and defensive strategies to help you use AI tools safely
We'll take a practical tour of threats, vectors, and defensive strategies to help you use AI tools safely
Intelligence Artificielle
AI is built on probability, not certainty. Security is built on definitely.
From image classifiers to Generative AI, these systems operate on “most likely correct,” raising new security questions. Can we apply the same deterministic models we’ve trusted for decades, or do probabilistic outcomes demand new thinking? This talk explores the risks, when traditional tools still apply, and how to keep teams safe as AI assistants reshape workflows.
From image classifiers to Generative AI, these systems operate on “most likely correct,” raising new security questions. Can we apply the same deterministic models we’ve trusted for decades, or do probabilistic outcomes demand new thinking? This talk explores the risks, when traditional tools still apply, and how to keep teams safe as AI assistants reshape workflows.
Intelligence Artificielle
Most Retrieval Augmented Generation (RAG) systems ship broken: embeddings without L2 normalization produce wildly inaccurate results. This talk shows why normalization matters, how to implement it in production (pgvector, TypeScript, Cloudflare Workers), and the massive accuracy gains it unlocks. Attendees will leave with practical patterns to avoid silent failure in their AI pipelines.
Intelligence Artificielle
Llama Stack is Meta's open-source framework designed to simplify the development and deployment of generative AI applications. It provides a consistent set of APIs regardless of the programming language you use (including Node.js, Python, and more). In this talk, we'll introduce you to the core concepts of Llama Stack and how it can help you build robust applications more efficiently.
Intelligence Artificielle
Cette conférence offre un guide pratique des vulnérabilités spécifiques à l'IA, techniques d'attaque et de défense, avec démonstrations de red teaming sur LLMs. Méthodologie d’attaque contrôlée (prompt injection, exfiltration), métriques de robustesse et mise en place de garde-fous runtime.
Intelligence Artificielle
Code reviews often get stuck on nits like spacing, naming, minor style. AI can automate the noise, freeing reviewers to focus on design, maintainability, and business impact. This talk shows how to leverage AI as a review partner, improving code quality and developer collaboration without losing the human touch.
Intelligence Artificielle
This session demonstrates how to detect, analyze, and prevent prompt leaks and persona failures in large language models. Participants will learn proven techniques and tools for securing AI prompts, enforcing consistent bot behaviour, and integrating evaluation and threat modelling into real-world engineering workflows.
Intelligence Artificielle
Learn how to build a modern development setup where coding agents can safely contribute code, run their own automated and manual tests, and debug bugs before handing changes back to you. We’ll cover dev containers setup, agent coding tools using MCP, and cross-agent pull request reviews. Attendees will see how to make agentic coding a reliable part of day-to-day development.
Intelligence Artificielle
As AI agents expand, scaling their ability to interact with external systems is key. Enter the Model Context Protocol (MCP)—a powerful design pattern connecting AI reasoning models with real-world tools. In this session, learn to build an MCP server in Node.js, letting AI clients query and interact with MySQL. We’ll cover architecture, integration tips, and share lessons learned along the way!
Intelligence Artificielle
Agents fail for two reasons: the model you're using isn’t good enough, or the context is a mess. Today, we focus on the mess. LLMs are pure functions, what goes in determines what comes out. That’s why context is everything. This session covers context engineering: how to write, select, compress, and isolate information. You’ll learn to avoid context rot, when to use memory, and how to split problems so models don’t drown in tokens.
Intelligence Artificielle
Building applications with AI and large language models (LLMs) adds a new layer of variability that makes testing difficult. Since interactions can be subtly different each time but still correct, traditional testing tools often fail. In this talk, we'll share our journey using the open-source tool deepeval to evaluate multi-turn conversations, tune prompts, and ensure our AI applications are reliable so that we could sleep at night.
Intelligence Artificielle
Last July, the Symfony AI was launched as an initiative, just like Symfony UX, to help developers integrate AI into their apps with ready-to-use components and bundles.
In this talk, let’s discover Symfony AI’s origins, explore its components, and learn how to use them with any AI platform. We’ll also cover how to get started using it in your PHP apps and contribute to the project!
In this talk, let’s discover Symfony AI’s origins, explore its components, and learn how to use them with any AI platform. We’ll also cover how to get started using it in your PHP apps and contribute to the project!
Intelligence Artificielle
Organizations struggle with surplus inventory disposal. Our startup tackles this with solar-powered smart containers using embedded Android devices that automate the reuse/recycle lifecycle.
This talk covers the technical journey: hardware selection, sensor integration, solar power challenges, edge ML for object recognition, offline-first architecture, and cloud sync strategies for distributed IoT deployment.
This talk covers the technical journey: hardware selection, sensor integration, solar power challenges, edge ML for object recognition, offline-first architecture, and cloud sync strategies for distributed IoT deployment.
Intelligence Artificielle
We're teaching AI to help humans sort surplus furniture using voice and vision, in real-time. Camera streams flow to AI models, voice streams flow back, decisions get logged.
The magic? WebRTC, without the traditional complexity! Discover how we built AI agents that feel like helpful coworkers: real-time video analysis, context-aware responses, and natural voice interactions. All while helping the environment.
The magic? WebRTC, without the traditional complexity! Discover how we built AI agents that feel like helpful coworkers: real-time video analysis, context-aware responses, and natural voice interactions. All while helping the environment.
Intelligence Artificielle
MCP is a new buzzword in 2025. However, should you always use MCP in your AI projects? The short answer is no. Therefore, let's take a look at the theory behind the Model Context Protocol and how to build a simple MCP client with Gemini, Claude and GPT. We will also go over some popular MCP servers such as Shopify's Storefront MCP.
Intelligence Artificielle
Everyone seems to be scared about AI taking over humanity. I don’t think so and I will present why in a Machine Learning perspective and importance of higher level consciousness. Fear is making us weak and doesn’t allow us to get a clear picture of the true challenge we are facing.
Intelligence Artificielle
FastAPI and Pydantic revolutionized API building in Python with strong typing, validation, and auto-docs. But generative AI often breaks that structure, messy JSON, brittle parsing, endless debugging. Pydantic AI brings the same rigor to AI agents: typed, validated, model-agnostic, observable. This talk shows how it works and how to use it to build reliable, structured AI Agent.
© 2010-2025 ConFoo. Tous droits réservés. Code de conduite
Gestion des Cookies
Pour offrir une bonne expérience, l'utilisation de cookies est nécessaire. Certains sont essentiels au bon fonctionnement du site, tandis que d'autres nous aident à mieux répondre à vos intérêts.