Dev Summit 2026

As LLMs become central to our workflows, the challenge shifts from "how to prompt" to "how to give AI secure access to our data" enter the Model Context Protocol (MCP).
In this session, we will demystify MCP servers and explain why they are the missing link in enterprise AI orchestration.
I will introduce the ADEO way of doing MCP servers with a toolkit designed to ensure that every AI tool we build remains compliant with our internal standards.
We will cover how to ensure all MCP servers implement OAuth 2.0, integrate seamless observability so teams can focus on adding value with MCP servers.
Learn how to empower AI agents while keeping ecosystem secure, observable, and standardized

We would like to talk about the observability of our physical stores.

There are many different types of devices in the stores: POS terminals, various scanners, displays, receipt printers connected to them, customer self-checkout stations, electronic queue terminals, printers, consultants’ computers, and so on. At the same time, physical stores also contain a large amount of network infrastructure — Wi-Fi access points, routers, firewalls, the physical network itself, and internet connectivity.

Unfortunately, devices sometimes fail, internet providers occasionally violate their service levels, and our services running in GCP can become unavailable for specific stores or specific business units due to new releases or changes in network or firewall configurations. In many cases, we learn about these issues quite late — only when customers encounter problems while paying for their purchases.

Last year, the Observability Adeo team worked very closely with the LMIT, LMPT, BROM, and LMFR teams to implement observability for this entire infrastructure. We designed an observability architecture for store devices based on OpenTelemetry, while the network team is implementing a solution for collecting network metrics based on Zabbix. Using data from both sources, we can build comprehensive dashboards that show the real state of the infrastructure inside our stores, allowing us to detect and respond to incidents in a timely manner.

We already have concrete implementations of this solution in LMIT and LMPT, and we would like to present them together in a joint presentation.

Following the successful launch of the LeaHER Tech initiative, we are pleased to take the next strategic step forward.
This session will be an opportunity to share the ambition driving us, as well as the roadmap that will guide our next steps.
We will focus on two key pillars of our development:
Global Deployment
An overview of the international initiatives supporting our mission, and how we plan to expand our impact beyond borders.
The LeaHER Tech Manifesto
The official unveiling of our manifesto — a document that defines our core values, our commitments to greater gender diversity in tech, and the action framework for all members of our community.
Join us to discover how, together, we are reshaping the role of women in the global technology ecosystem.

The "Chat" era is dead. In 2026, the real work happens in the Terminal. Stop the disposable apps and the AI-induced FOMO; it's time to reclaim your craft.

In a world of infinite code generation, Architecture and Maintainability are the only things that separate a resilient product from an unmaintainable liability. We have reached a point where generating code is easy, but controlling it is the true engineering challenge.

Join this 90-minute hands-on lab to cut the noise, kill the "vibe coding," and master the complexity of production-ready software. Step out of the chatbox and learn how to orchestrate true Agentic Workflows.


# Lab Details: 90 Minutes to Deep Dive into 2026 Engineering

This is not a session about writing better prompts; it’s a session about upgrading your entire operating model as a developer. We will build, debug, and secure a complex application by mastering four pillars:

**1\. The Command Center (CLI First)**

Transform your CLI into an autonomous powerhouse. We will move away from passive IDE sidebars and demonstrate how to trigger agents directly from the terminal to execute complex refactoring, run diagnostics, and manage infrastructure in real-time.

**2\. Skills over Chat (Context Engineering)**

Stop prompting, start building. You will learn how to give your environment "Capabilities" (Skills and Context) to automate the grind. By injecting architectural rules and domain constraints directly into the workspace, you allow the agents to focus on your *intent* rather than forcing you to babysit their output.

**3\. The Shadow Team (Agentic Orchestration)**

You are no longer just a coder; you are the Lead Engineer of a digital squad. You will orchestrate specialized AI personas—a Security Auditor, a QA Automator, and an Architecture Reviewer—that act autonomously to challenge your plans, generate E2E tests, and secure your Pull Requests.

**4\. Engineering vs. Tokens (Sustainable Craftsmanship)**

AI can generate a million tokens per minute, but who maintains them? We will focus on the ultimate 2026 skill: System Design. You will learn how to enforce boundaries and review Agent Plans *before* they touch the codebase, ensuring that AI speed never compromises software excellence.


# Participant Prerequisites

* Professional software engineering experience (any modern stack).
* A laptop with a modern IDE and terminal access.
* The mindset of a Tech Lead willing to stop coding everything manually and start orchestrating.

Scaling complex AI deployments demands overcoming severe latency and hardware bottlenecks, especially for resource-intensive GenAI and Deep Learning models. This session dives into the OpenVINO toolkit to demonstrate how you can seamlessly bridge the gap between trained models and high-performance production environments. You will learn practical, code-level strategies to convert, optimize, and accelerate inference for both generative and conventional AI models across today's most popular frameworks.

A broken release on an e-commerce platform can have immediate and important financial impacts. At Decathlon, with dozens of workloads each deployed up to several times a day, the risk of human error-related incident is high.

In that feedback session, we'll share how Flagger enabled us to match velocity and serenity. By automating progressive deploy strategies, we secured the change process : less impacts from unavoidable incidents and more confident developers.

You'll discover our wins, but also the main challenges we faced and overcame along the road.
Expect a comprehensive description of the process with concrete examples, enabling the same implementation, however your kubernetes clusters are configured.

What if the experience our customers have with the best in-store advisors could be available anytime, anywhere?

In this talk, we share how we are designing a conversational AI platform that extends the expertise of retail teams beyond the store, 24/7, to instantly connect customers to the right information, advice, and services.
We'll explain why we approach conversational AI not as a simple chatbot, but as a platform connecting expert AI assistants, business tools, and customer journeys.

We'll also share organizational challenges we faced when designing conversational experiences that actually help.

This session introduces Agent Skills, a standard designed to specialize AI tools like Copilot using "skills" that combine specific instructions and contextual resources. Through live demonstrations, participants will learn how to create and share these skills to transform a general-purpose AI into a truly adaptable and collaborative business assistant.

During this session, we will also highlight some key skills from Adeo and Decathlon

Decathlon's Data Lake is organized into progressive layers that transform data through increasing levels of complexity to power reporting, visualization (e.g., on interactive dashboards), and eventually advanced Machine Learning & AI (e.g., product recommendation, demand forecasting, dynamic pricing, …). To achieve this, we build and maintain complex, distributed pipelines written in SQL, and we leverage Apache Spark’s engine to handle Big Data processing at scale on multi-node clusters. However, complexity comes at a cost: as we stack more and more data transformations, manually tracing the exact origin of a specific data item becomes increasingly difficult and unmanageable, creating a critical need for an automated solution. We have recently recruited a Data Engineer intern and partnered with academic experts from ENS - PSL and Université Grenoble Alpes to prototype a (Fine-Grained) Data Provenance tool compatible with Apache Spark.
The ability to track the provenance/lineage of granular data portions is critical for:
- Trust & Reliability: guaranteeing the accuracy of results for data consumers.
- Root Cause Analysis: diagnosing anomalies (e.g., aberrant turnover figures) to pinpoint the exact source of a problem.
- Impact Analysis: predicting how data updates will propagate through our versioned datasets.
- GDPR compliance: ensuring that sensitive data (PII) does not unintentionally "leak" into refined datasets.
- Testing: extracting representative subsets of data for lightweight integration tests and prototyping.
In this talk, we will present few concepts of data provenance and present where we currently stand and what we plan to build in the future.

## En Français

Titre : La sécurité en double aveugle : venez relever le défi du kata d’architecture et cybersécurité communautaire

Quand on définit un système, est-ce qu’il vaut mieux réfléchir d’abord à l’architecture ou à la sécurité de ce système ? Vous avez une opinion ? Pas nous ! Venez participer à cet atelier, dans lequel nous vous ferons réaliser une succession de katas pour déterminer ensemble quel ordre est préférable, si tant est qu’il y en ait un.

Nous verrons dans ces katas comment intégrer correctement le raisonnement sur la sécurité à la conception de ce système. Nous parlerons évidemment des utilisateurs légitimes de ce système et de leurs besoins de sécurité, mais aussi des utilisateurs moins légitimes et de la façon correcte d’imaginer leurs interactions avec ce système. Pour cela, vous marierez les méthodes d’architecture agile et celles de définition de risque.

Vous ressortirez de cet atelier avec des idées plus claires sur les manières d’avoir des conversations correctes avec les différentes parties prenantes de votre projet en ne mettant ni l’architecture ni la sécurité de côté. Et peut-être avec un bon ordonnancement des questions d’architecture et de sécurité si nous trouvons un consensus.

**Prérequis pour l'atelier** : N'amenez pas votre ordinateur, il ne servira à rien. En revanche, amenez votre capacité de réflexion, elle sera mise en oeuvre.

## In English

When defining a system, is it better to think about the architecture first, or the security? You have an opinion? We don’t! Come join us for this workshop, where we will lead you through a series of katas to determine together which order is preferable—if there even is one.

Throughout these katas, we will explore how to properly integrate security reasoning into the system's design. We will obviously discuss legitimate users and their security needs, but also less legitimate users and the correct way to envision their interactions with the system. To achieve this, you will marry agile architecture methods with risk definition techniques.

You will leave this workshop with clearer ideas on how to have productive conversations with your project’s stakeholders without sidelining either architecture or security. And perhaps, if we reach a consensus, we'll walk away with a solid sequencing for architecture and security questions.

**Workshop Prerequisites**: Do not bring your computer; it will be of no use. Instead, bring your thinking caps—they will be put to work.

Server-Driven UI: In today’s competitive landscape, agility is often hampered by the silos between Web and Mobile development. While Web teams can deploy in minutes, mobile teams have traditionally been limited by two-week sprint cycles and unpredictable Play Store / App Store approvals.
This friction prevents Business Units (BUs) from executing truly synchronized, multi-channel marketing strategies with native experience.

What if we could offer a "One Gesture" solution to our BUs?

Imagine giving the keys to the app directly to the BUs. That's exactly what we did. With ContentBuilder, our Server-Driven UI platform, BUs can now create, test, and publish complex landing pages and marketing content in real-time in a single step.

In this talk, we will demonstrate how we broke these silos by building ContentBuilder, a Server-Driven UI platform. This strategic shift moved the power of UI creation from core engineering teams directly to the BUs. We will show how a single configuration allows them to build, test, and ship complex landing pages and marketing content simultaneously across Web and Mobile in real-time.

Key Takeaways:
- The Multi-Channel Driver: How the need for instant, synchronized content pages reactivity across all touchpoints (loyalty, xxx) dictated our roadmap.
Democratizing Design: How we built a unified platform that allows BUs to assemble pages using a shared library of pre-validated components, ensuring brand consistency without a single line of code through ContentFull.
- The Architecture of Autonomy: A deep dive into the JSON-based contract that bridges our backend with both Web and Flutter environments, mirroring a truly universal Design System.
- Measuring Impact: Moving from a 20-day release cycle to a 5-minute "Go-to-Market" and the resulting uplift in cross-platform user engagement.
- Hard Truths & Trade-offs: The challenges of cross-platform versioning, maintaining native performance on mobile vs. responsiveness on web, and ensuring stability.

Let’s dive together into the journey of one of Decathlon’s strategic initiatives: scale circularity, moving it from “new business models” to “business as usual”.

CONTEXT
Over the last 18 months, we’ve been exploring how to “natively” integrate circularity (second life, rental & repair) into Decathlon’s new eCommerce platform (revamp), used by millions of customers across our core markets (France, Spain, Italy, etc.)
-> Since day 1, the team’s mission has been simple: leverage eCommerce to scale circularity’s adoption and business. 

OUR MAIN CHALLENGE
Building embedded solutions that deliver business value for Circularity, without putting Decathlon’s core business at risk (aka selling new products)

OUR CONVICTION
Enabling customers to discover, consider and choose circular alternatives in a natural and transparent way… Aligned with Decathlon’s philosophy of “libre meilleur choix” (“free best choice”)

THEMES WE WILL COVER
Product discovery, multiple stakeholder engagement, driving outcomes vs building outputs, co-design, experimentation culture (A/B testing), shaping new consumption behaviors, gaps between what people say vs what they do, etc.

SPOILER ALERT 🚨
We will be sharing big wins, but also epic fails :)

Together, let’s “Make Circularity Great Again the New Normal!” ✌️



[FR version below]

"Circularity as business as usual" - intégrer de façon native les offres de l'économie circulaire au sein du eCommerce chez Decathlon

Je vous propose de partager un retour d'expérience sur le travail mené depuis 1 an et demi par mon équipe sur l'intégration de l'économie circulaire (seconde vie, location et réparation de produits sportifs) au sein du site eCommerce de Decathlon (la nouvelle plateforme "revamp", aujourd'hui déployée dans tous les principaux pays de Decathlon et utilisée par des millions de clients en France, Espagne, Italie, Allemagne, Belgique, Pays Bas, etc.)

Le challenge porté par l'équipe était simple: faire progresser la part du digital sur le business de l'économie circulaire, en se concentrant particulièrement sur l'amélioration du trafic et de la conversion. Notre parti-pris: intégrer ces différentes offres de la façon la plus naturelle et transparente possible, de façon à mieux faire connaître ces offres alternatives, tout en assurant le "libre meilleur choix" aux clients de Decathlon.

Spoiler alert: on a eu de beaux succès, mais aussi des "epic fails" (car tout ne s'est pas passé comme prévu)

Au sein de ce vaste sujet, voici autant de sous-thématiques qu'il sera possible d'aborder (avec plus ou moins de profondeur): product discovery, gestion des parties prenantes (sujet au carrefour de multiples équipes), outcome vs output, co-design, culture de l'expérimentation (A/B testing), transformation des habitudes de consommation, écarts entre les tests utilisateurs (ce que les gens disent) et la vie réelle (ce que les gens font), etc.

Offering the right product in the right place: the Holy Grail of e-commerce. Yet, in a catalog of millions of items, a bestseller can become "noise" if it appears out of context. While popularity has long dominated ranking systems, it has now hit a glass ceiling. How do we move from purely statistical sorting to more intelligent navigation?

Within the AAAI team at ADEO, we reached a radical conclusion: behavioral signals (clicks, sales) modeled by classic ML are no longer enough. They are semantically blind. The result? A category can become flooded with accessories or unrelated products simply because they are cheaper and highly clicked, ultimately breaking the browsing experience.

In this session, we will take you behind the scenes of our hybrid ranking architecture. We will detail how we integrated the power of Transformers into the heart of our engine to re-inject context where algorithms were once "deaf" to the meaning of words.

On the Agenda:
The End of "Click Dictatorship": Why behavioral ML alone creates harmful popularity bias.

Semantic AI in Action: How we use Transformers to map user navigation intent to catalog reality.

Hybrid Architecture: Making the precision of a Cross-Encoder coexist with the scalability of a ranking model (like XGBoost).

Results & Lessons Learned: Real gains measured in ranking quality, limitations encountered, and takeaways from deploying on a massive scale.

Techniques covered: Python, ZenML, Transformer Models, Statistics, NDCG, ML

Target Audience: Data Scientists, ML Engineers, Product Owners

Level: Intermediate

At Pellenc ST, artificial intelligence is deployed directly into industrial systems operating in real-world waste sorting environments. This talk shares practical feedback on transforming AI prototypes into reliable, production-ready solutions with measurable ROI. We will present concrete use cases combining computer vision and spectrometry, and discuss the challenges of working with complex, noisy data collected under harsh industrial conditions. The session will cover key topics such as model robustness, edge deployment constraints, and system integration within high-performance machines. A strong focus will be placed on operational excellence: monitoring model drift, ensuring observability, and maintaining performance over time. We will also share insights on cost-performance trade-offs and technology choices in an industrial context. Finally, we will highlight how AI can contribute to sustainability by improving sorting efficiency and material recovery — demonstrating that AI can be both economically and environmentally impactful.

Decathlon's decentralized analytics organization faced **KPI Chaos**, which was a bottleneck for next-gen BI and **AI-readiness**. This talk explores how establishing an **Analytics Center of Excellence (CoE)** enabled the delivery of a **federated Semantic layer**.

The Analytics CoE is driven by **Staff Engineers for Analytics**. Using **transversal leadership** and a **communities-based model**, we were able to source bottom-up bandwidth for transversal projects without dedicated budgets or direct hierarchical authority.

This empowered the delivery of `insight-all-metrics`, a **federated Semantic layer** built on **Databricks Metric Views** and **dbt**. Domains centrally expose their shared KPIs via simple YAML files. Then, any data practitioner can easily compute those KPIs. Leveraging **materialization features** reduced compute costs and enabled the instantaneous querying required for real-time GenAI applications.

By gathering together the **Metric Layer** (SQL logic to compute KPIs) and the **Semantic Layer** (natural language context) within a single repository, we bridged the gap between human questions and database queries. This single governed source of truth now powers both **Tableau Pulse** and **Databricks Genie**, marking our successful transition to **GenAI for BI**.

I will live demo some features.

In a large-scale tech organization sending an email seems like a solved problem – until you look under the hood and discover that teams keep reinventing the wheel. This leads to massive code duplication, fragmented implementation and lack of centralized observability.

In LMIT we decided to tackle this technical debt by building a reusable library designed as a high-impact MVP. This session is a “return of experience” on the journey of creating a centralized deployment model within our corporate infrastructure, starting from BU Italy with the ambition to scale. We'll dive into the architectural challenges of creating a tool that is robust enough for enterprise scale yet simple enough for developer adoption.

We'll discuss
- How a shared library reduces boilerplate and speeds up the "time-to-market" for new features.
- Our architectural approach to ensure security, traceability, and high deliverability by design, moving away from "every team for themselves".
- Strategies to make the library easy to use, focusing on intuitive APIs, and seamless integration into existing developer workflows.
- The challenges of integrating a common tool into the global infrastructure while maintaining BU autonomy.

This presentation will explore how we address the complexities of network management in large-scale environments. In such a landscape, responsibilities are shared between the central and local teams. The central team is in charge of the SD-WAN solution, including providing training and advice to local teams, while the local teams are responsible for their own specific perimeters, which can be one or multiple sites.

A primary goal for us was to ensure that people can manage their perimeter in a simplified and straightforward way, free from excessive complexity.

In this sense, this interface—or portal, as we call it—was a direct response to the challenge of managing a growing number of sites (stores, warehouses, headquarters, etc.). It is designed to simplify the management of SD-WAN resources like firewall rules and devices.

Ultimately, the portal aims to transform Decathlon's network management from a manual, site-by-site process into a more scalable and automated approach.

What You Will Learn
- What was the initial motivation for creating this portal? And why did we create it?
- How do we monitor the management of the sites?
- How do we prevent users from making unauthorized configuration changes?
- The process for creating new Decathlon sites from an SD-WAN perspective and how local teams can manage their own sites.

Our Product Group manages core reference data for products and partners used across Retail supply chain operations. This foundation supports critical processes including sourcing, planning, distribution, and wholesale execution.

In this session, we will share our experience integrating Generative AI into an established and complex product ecosystem.

Our discussion will focus on:

-Embedding GenAI into our Group Product Strategy

-Driving AI adoption through internal AI champions

-Creating an innovation space within a complex, enterprise-scale environment

-Integrating GenAI into non-native digital products

-Breaking organizational and system silos, particularly within an SAP-based ERP landscape

-Building the Operating Model progressively while delivering tangible business value

We will also present a concrete innovation use case:
a GenAI-powered solution supporting Business Partner creation for Wholesale operations, illustrating how AI can streamline processes, reduce friction, and generate measurable operational impact.

This session highlights the reality of deploying GenAI in a structured enterprise context — not as a standalone experiment, but as an embedded capability within core business systems and operating models.

Coding Agentic Systems is fun — it feels like magic!
Vibe-coding Agentic System is even more fun — double magic, hah!?

But quality evaluation of LLM generations? That part is usually… boring.
And if you’re not in the Python ecosystem, good luck finding a framework that actually works for you.
So eval tasks quietly sit in our backlogs, waiting for better days, the v2 release, or some future “we’ll fix it later.”

Drawing from real experience building production RAGs, agentic pipelines, and LLM-powered features across different stacks, I'll share the lessons learned the hard way — what broke, what worked, and what we wished we'd measured from day one.
The goal of this talk is simple: make evals understandable, and the creation process easy — with Coding Agents doing the heavy lifting alongside you. Practical tips, tricks, and a fresh perspective on making evaluation a natural part of developing LLM-powered applications.

In a modern "RUN" environment, the challenge isn't a lack of data, but the time required to make sense of it. Manually auditing an incident to understand why an alert didn't fire—or identifying recurring patterns across hundreds of tickets—is a tedious process that often takes hours of manual "investigative" work.

At LMFR, we’ve automated these critical tasks using Agentic AI and the Model Context Protocol (MCP). By connecting AI agents directly to our monitoring and ticketing stack (ServiceNow, PagerDuty, Datadog), we have transformed post-incident analysis into a 10-minute automated process.

In this 30-minute demo-driven session, we will deep-dive into two major use cases:

- The Automated Observability Audit: We will demonstrate how an AI agent analyzes a past incident to answer three crucial questions: What was missing to prevent this? What could have detected it faster? How can we strengthen our alerting? The agent cross-references logs and traces to identify the "invisible" gaps in your monitoring strategy, ensuring the same incident doesn't catch you off guard twice.

- Smart Incidentology & Trend Analysis: Moving beyond individual tickets, we’ll show how AI can analyze the entire history of a product to uncover structural flaws. By identifying the most frequent defects and recurring pain points, the agent helps Ops teams move from "Firefighting" to true Problem Management, providing a data-backed list of priorities for the product backlog.

What if architectural artifacts allowed you to see things from new perspectives, shedding new light on making more informed decisions? Yes, but architecture is complex, and I don't understand all the concepts... What if it were now accessible and user-friendly?

We've spent years writing SQL, building Python pipelines, and waiting weeks for insights stuck behind a Jira ticket. What if anyone could go from a question to an answer with just a prompt?

Let's be honest, until late 2025, that was fiction. The demos looked great on stage. Production? Not so much. But something actually shifted. Models got genuinely good at structured output, MCP became the standard to connect LLMs to real tools, and DuckDB made it possible to query anything without infrastructure.

In this talk, I won't oversell. I'll build a full analytics workflow live — from raw data to insights using just natural language. First with a frontier model to show what's actually possible today with DuckDB and MotherDuck. Then with open-source models running entirely locally — no data leaving your network, no cloud dependency.

If you've been burned by AI demos that only work on slides, this one's for you. Expect a real demo, honest tradeoffs.

Join three Global Transformation Success experts for a 30-minute deep dive into the world of Maturity Models (Project & Product Management). We will strip away the bad reputation of these assessments and reveal how they serve as powerful catalysts for standardization, continuous improvement, gain in reliability for Digital Products, and help to industrialise their deployment.
Thanks to concrete examples in Adeo Ecosystem (Global Ready, Engineering Index, Adeo Product Grade, Activation Maturity Index, UX Bar, Change Ready,..), we will identify some tips to design, execute and animate these frameworks.

The "Chat" era is dead. In 2026, the real work happens in the Terminal. Stop the disposable apps and the AI-induced FOMO; it's time to reclaim your craft.

In a world of infinite code generation, Architecture and Maintainability are the only things that separate a resilient product from an unmaintainable liability. We have reached a point where generating code is easy, but controlling it is the true engineering challenge.

Join this 90-minute hands-on lab to cut the noise, kill the "vibe coding," and master the complexity of production-ready software. Step out of the chatbox and learn how to orchestrate true Agentic Workflows.


# Lab Details: 90 Minutes to Deep Dive into 2026 Engineering

This is not a session about writing better prompts; it’s a session about upgrading your entire operating model as a developer. We will build, debug, and secure a complex application by mastering four pillars:

**1\. The Command Center (CLI First)**

Transform your CLI into an autonomous powerhouse. We will move away from passive IDE sidebars and demonstrate how to trigger agents directly from the terminal to execute complex refactoring, run diagnostics, and manage infrastructure in real-time.

**2\. Skills over Chat (Context Engineering)**

Stop prompting, start building. You will learn how to give your environment "Capabilities" (Skills and Context) to automate the grind. By injecting architectural rules and domain constraints directly into the workspace, you allow the agents to focus on your *intent* rather than forcing you to babysit their output.

**3\. The Shadow Team (Agentic Orchestration)**

You are no longer just a coder; you are the Lead Engineer of a digital squad. You will orchestrate specialized AI personas—a Security Auditor, a QA Automator, and an Architecture Reviewer—that act autonomously to challenge your plans, generate E2E tests, and secure your Pull Requests.

**4\. Engineering vs. Tokens (Sustainable Craftsmanship)**

AI can generate a million tokens per minute, but who maintains them? We will focus on the ultimate 2026 skill: System Design. You will learn how to enforce boundaries and review Agent Plans *before* they touch the codebase, ensuring that AI speed never compromises software excellence.


# Participant Prerequisites

* Professional software engineering experience (any modern stack).
* A laptop with a modern IDE and terminal access.
* The mindset of a Tech Lead willing to stop coding everything manually and start orchestrating.

When a critical incident strikes a complex system, every second counts.

The initial moments—the "time to engage"—are often lost trying to identify and gather the right experts. This delay can impact our customers and our business.
Who to call? How long does it take to get the right experts in the same room? Should we communicate among our Business Units ?

At ADEO, we have evolved our incident management process. Over the past year, we have implemented a "War Room" system for Priority 1 incidents on Google Chat.

With a single trigger, this tool automatically:
- Identifies and mobilizes the correct product and operations teams.
- Creates a dedicated Google Chat space and Google Meet for immediate collaboration.
- Provides incident managers with the tools they need to lead effectively and accurately.

In a 30-minute session, we can present this innovative tool, its key features, and the significant impact it had on reducing our time to engage and improving transparency.

Legacy codebases often sit in valuable but neglected state because migrations are costly, risky, and domain-knowledge intensive.

Large language models help change that. By analyzing code, tests, and docs, LLMs can produce helpful code changes and findings in seemingly forgotten parts of our legacy code. As a force multiplier, LLMs can help accelerate migrations to modern stacks and reduce the cognitive overhead, letting small engineering teams execute bigger migrations with larger confidence.

In this talk we'll go through:
- LMPL's experiences with migrating legacy PHP code to modern Java stack
- efficiency gain within a small and focused migration team
- how to counteract the traps and issues with LLM-driven code migrations of large codebases