Intelligence has become free.
Discernment has become a luxury.

Jérôme Roussin
AI Speaker & Local AI Expert

AI is transforming society and business.
But closed AI has a hidden price: dependency, exploding bills, data leaving your control.

💻 Local AI & open source expert — conferences, audits and deployments for businesses that want to control their costs and their data.

🎯 Request a conference See the 2026 assessment
300+ Conferences
10 to 3000 Attendees
2 TEDx
10 Countries
4.7/5 Client Satisfaction
📊 The 2026 Assessment

Companies have burned through their AI budgets in 2026 in four months.

300k€

Annual closed-cloud cost · 50 users

500 € par mois et par utilisateur. Uber et ServiceNow burned through their annual budget in four months. Providers have changed their pricing model. Bills have tripled.

AI agents multiply requests by 10, by 100. Google processes 3.2 million billion tokens per month — seven times more than a year ago. Providers can cut access overnight, as Anthropic did by government decree in June 2026. Your data goes to servers you don't control.

Arthur Mensch, CEO of Mistral AI, told French lawmakers: "an inability of actors to disengage from these new services." Lock-in is a current reality, and really not a future risk! It's happening now, with every integration.

The alternative exists. An open source model, served by Ollama, running on a Mac, on a server, in your company. Fixed cost, unlimited usage, local data. The field is proving me right faster than I expected. It's not ideology — it's economics.

🎯 FLAGSHIP AI CONFERENCES

My three signature conferences

Why and how to move away from closed cloud models toward local AI, open source and on-premise. The assessment, the math, and the deployment.

🛡️
01
SOVEREIGNTY

Breaking Free from Closed Cloud Models

AI bills are exploding. Closed-model providers can cut access overnight. Your data leaves your control. This conference presents the economic assessment and shows how local and open source AI — with Ollama — lets you take back control: fixed costs, local data, zero lock-in. From closed cloud to open source, local and on-premise.

🛡️ What participants will learn:

  • The math — Closed cloud vs local in euros: €300k vs €20k for 50 users
  • Lock-in — Why dependency on closed cloud models is the new vendor lock-in, but worse
  • Sovereignty — Anthropic cut by decree: sovereignty starts with the server in your basement
  • Practical guide — How to get started with Ollama in three commands, on Mac or server

⏱️ Available formats

90 min · Half-day · Full-day

👥 Ideal audience

CIO, CFO, Executives

🤝
02
AWARENESS

Human-AI Orchestration

AI is neither a threat nor an extension. This conference explores how to think with AI when it becomes an extension of your cognition, and how to orchestrate human-AI systems to create more value. From initial competition to hybrid collaboration.

🤖 What participants will learn:

  • Hybrid Thinking — How to think with AI as a cognitive extension
  • Focus Map — Where human involvement IS mandatory vs desirable
  • Orchestration — How to arrange systems for intelligent hybridization
  • Collaboration — Transforming the human-machine relationship into partnership

⏱️ Typical Duration

90 minutes

👥 Ideal audience

CIO, CTO, Transformation

📊
03
VISION

The End of the Knowledge Economy

The knowledge economy is dead. Value has migrated to what AI cannot replace: orchestration, vision, and discernment. This conference reveals the new levers of competitiveness in a world where intelligence has become free.

📚 What participants will learn:

  • Allocation — How to optimally allocate resources between humans and AI
  • Arrangement — The art of arranging systems and processes to create more value
  • Responsibility — Who takes responsibility when AI predicts?
  • Discernment — The critical skill becoming the new luxury

⏱️ Typical Duration

90 minutes

👥 Ideal audience

C-Suite, Executive Committees

💡 Available themes: All three themes can be customized to your sector, challenges, and audience.

🎯 INTERVENTION FORMATS

Moving to local AI

Conferences, workshops and deployments to move away from closed cloud models toward local and on-premise AI.

🖥️

Local AI & On-Premise Workshop

Hands-on workshops to deploy open source models with Ollama on Mac or on-premise servers. Installation, model selection, optimization, agent deployment. The differentiating workshop: no generalist offers this. Hands-on and immediately applicable.

🎤

Conferences & Keynotes

Interventions that enlighten and inspire! Conferences on AI tools in business, data sovereignty, local AI as a viable economic alternative, and AI's impact on your business.

🎓

Custom Training

Educational programs to train your teams with follow-up and recurrence. Content adapted to your sector and specific needs. From beginner to on-premise deployment.

⏱️ DURATIONS

Flexible formats

From a 90-minute conference to a full 2-day deployment.

45
Minutes

Mini conference

90
Minutes

Standard conference

1/2
Full day

with Workshops

1
Full day

with Workshops

2
Full days

Custom

Options

Formats: for 45 and 90 minutes, the talk can be generic on the local AI assessment, or customized to the client company's market and specific needs.

English conferences: available (all formats)

Pricing: available on request — contact me for a personalized quote.

📰 Articles & Analysis

Local AI in practice

Analysis and field reports: AI costs, sovereignty, Ollama, on-premise deployment. The assessment, the math, and the field.

Costs Closed Cloud

The cheapest AI is the one you host yourself

€500 per month per user in the cloud. €20,000 once for a server. The gap is an order of magnitude, and really not just a margin. The math nobody does.

Published June 20, 2026 · Read article → · 5 min read

Local AI Ollama

How to get started with local AI in three commands

Ollama, an open source model, a Mac or a server. No account, no license, no cloud. The minimum viable guide to deploy on-premise AI.

Published June 20, 2026 · Read article → · 4 min read

Sovereignty Geopolitics

Anthropic cut off: when Washington takes back control

In June 2026, the US administration banned foreign access to Anthropic's models. A European client woke up to an indispensable tool that no longer worked. Sovereignty starts with the server in your basement.

Published June 20, 2026 · Read article → · 6 min read

Agents Costs

The local agent paradox

AI agents multiply requests by 100. In the cloud, every action is billed. Locally, the marginal cost is zero. Agents make local AI economically obvious.

Published June 20, 2026 · Read article → · 4 min read

Open Source LeCun

Yann LeCun was right about one thing nobody remembered

"Foundation models will become an infrastructure of the information society." Infrastructure like electricity. We didn't let an oligopoly control that.

Published June 20, 2026 · Read article → · 3 min read

Lock-in Lock-in

AI dependency is the new Microsoft lock-in, but worse

We've seen this movie with Office, Exchange, Active Directory. Except with AI it's worse: your data contextualized their models, your workflows depend on their API, and migration costs grow over time.

Published June 20, 2026 · Read article → · 5 min read

🧠 Questions to ask yourself

The questions nobody asks

The ones that stay in blind spots. The ones that make the difference between a controlled bill and an expensive dependency.

What happens if my AI provider cuts access tomorrow? +

In June 2026, the US administration banned foreign access to Anthropic's models. Overnight. If all your AI depends on a cloud API, you're at the mercy of a decision you don't control. A local model, on your server, cannot be cut off.

How much does cloud AI really cost when you deploy agents? +

A chatbot is one request and one response. An agent is 10, 50, 100 requests to accomplish a task. Each step is billed. Uber and ServiceNow burned through their annual budget in four months. The marginal cost of a local agent: zero. The marginal cost of a cloud agent: every action.

Where exactly do my confidential data go? +

When your employees paste content into a cloud AI assistant, your internal documents, your source code, your client data goes to servers you don't control. GDPR formally requires consent for transfers outside the EU. In practice, thousands of employees do it every day without anyone knowing where it ends up. A local model: your data never leaves your network.

Is an open source model really usable in business? +

Yes. Mistral, Llama, Qwen are production models used by companies worldwide. They don't match GPT-4 on everything, but for most use cases: summarization, extraction, classification, rephrasing, code... they're more than sufficient. And they run with Ollama in three commands. The required hardware: an Apple Silicon Mac or a server with a GPU.

How long does it take to move from cloud to local? +

Installing Ollama takes 5 minutes. Downloading a model takes 10 minutes. The first local request takes 30 seconds. For individual use on Mac, it's an afternoon. For an on-premise deployment for 50 users with a dedicated server, it's 1 to 2 days. It's not a 6-month project. It's a few days.

What if we've already invested in cloud APIs? +

Local AI doesn't systematically replace the cloud. The right model is hybrid: cloud for tasks that need the latest frontier, local for recurring, internal, large-scale usage. An AI usage audit identifies what should stay in the cloud and what should move local. Le gain se fait sur les volumes, pas sur tout.

Is local AI an ideological or economic position? +

Economic. An unlimited variable cost always ends up exceeding a fixed cost. €300,000 per year in cloud vs €20,000 once for local, for 50 users. The gap is an order of magnitude, and really not an opinion. Open source is not a developer ideology: it's the technical condition for your company's information infrastructure not to be privatized end to end.

Do any of these questions resonate?

Don't leave the answer in the blind spot. A 20-minute discovery call is enough to assess where you stand and identify the first levers.

🎯 Book a discovery call
WAVE Principle
🌊 WAKE • AWARENESS • VISION • EMPOWERMENT

The WAVE method to understand today's AI

WAVE is my framework to understand today's AI: awaken consciousness to the urgency, understand the real impacts, build a strategic vision, and give the keys to act with local and on-premise AI.

W

WAKE UP — Awaken consciousness to the AI urgency

The first step is recognizing that AI is not a technological wave that will pass. It's a fundamental paradigm shift.

⚠️ What Wake implies:

  • Recognize critical weak signals
  • Understand competitors' insurmountable advantage
  • Accept that procrastination is no longer an option

💡 Diagnostic: Wake helps you identify where your organization stands and what the strategic blind spots are.

A

AWARENESS — Understand human and ethical impacts

We must understand AI's real impact on humans, ethics, and organizational structure. It's not about technology, but about humanity.

🔍 What Awareness implies:

  • Understand Moravec's paradox
  • Distinguish what can be automated vs what's human
  • Identify ethical consequences

💡 Perception: Awareness develops hybrid perception — thinking with AI as a cognitive extension.

V

VISION — Build a strategic perspective

With consciousness and understanding in place, we must build a vision of the future. Not a technological vision, but a strategic one in the post-knowledge era.

🎯 What Vision implies:

  • Define where to navigate: allocation, arrangement, responsibility
  • Identify strategic weak signals
  • Build a transition roadmap

💡 Direction: Vision gives strategic focus capacity — from reactivity to foresight.

E

EMPOWERMENT — Give the keys to act

Understanding is not enough. Empowerment gives you the tools, methods, and confidence to transform your organization.

🚀 What Empowerment implies:

  • Implement human-AI system orchestration
  • Develop organizational "taste"
  • Create intelligent hybridization structures

💡 Action : Empowerment gives you concrete keys — intervention formats, training, custom support.

💬 Testimonials

What my clients say

Business leaders and opinion leaders testify to the quality of my interventions on AI in business.

"Jérôme shows remarkable pedagogy in explaining complex concepts, with very concrete and detailed examples."

Stanislas Desjonqueres

PDG, BIOSE

"Revolutionary and exceptional content. One of the best encounters. It can challenge many things."

Elena Lardy

Cofondatrice, TSAR VOYAGES

"Thank you for demystifying this fascinating topic with clarity, and for identifying the stakes for businesses."

Jean-Luc Wilain

Business Director, IRELAND BLYTH

"Excellent intervention, highly pedagogical on a complex subject mixing many different considerations."

Manuel Davy

PDG, VEKIA

"A most complex subject, and you managed to make it understandable and fascinating."

Francois Aubery

PDG, MAVN / ALUPRO

"By far the best intervention I've attended on the subject. Very impactful presentation!"

Carole Catry

General Secretary, RABOT DUTILLEUL

"Bravo for the WAVE method! Thank you for the effective pedagogy. I finally understand..."

Hervé Le Gavrian

Business Director, MICHELIN

🤝 Clients & Partners

They trust me

Ready to explore local AI options?

🎯 Conferences, AI cost audits, local and on-premise AI workshops, Ollama deployment. Local AI is a serious alternative for the future of your business.

🎯 Request a conference
📬 Contact

Let's discuss your move to local AI

Jérôme Roussin

Book Jérôme for a conference or local AI workshop

AI conferences, local & on-premise AI workshops, AI cost audits, Ollama deployment. Tell me where you stand: cloud prototype, exploding bill, or ready to go local?

👥 LinkedIn : visit my profile
📍 Europe & International
🎤 Conferences in French and English
🔍 Conferences, AI cost audits & on-premise Ollama deployments

🔒 Your data is processed in Germany (GDPR)