Who we are, where these five sessions go, and what you walk out with today.
Five sessions
Session 1The foundations
Session 2Your own examples
Session 3Prompting in practice
Session 4Tools in your workflow
Session 5Building your habits
Everything we do in the next four sessions builds on today.
Your two speakers
Olivier Durand
Co-founder, ADALAN
Etienne Galland
Co-founder, ADALAN
AI only creates a real advantage when teams change how they work. That is why you are in this room.
This is not a technical training
You will not write code.
You will not need to remember any formula.
You will get a mental model: a picture in your head of what happens when you type something into one of these tools.
Once you have that picture, you stop being a passenger.
02
Generative AI and LLMs
What these tools are, and the single idea everything else rests on.
Generative AI & LLM
It creates. And it predicts.
Generative AI
Before, AI sorted things. Spam, or fraud.
This one creates. Text, images, audio, code.
LLM
Large Language Model. The one that creates text.
It predicts the next word. It is not looking anything up.
Picture a colleague who has read almost everything, up to a date. Very confident. Always an opinion. And three limits.
No access to your CRMNothing about your clients.
No access to your filesNever seen a document of yours.
Stopped reading at a dateNothing about last month.
03
Tokens and the context window
What it actually reads, and how much of it fits at once.
Tokens
It does not read words. It reads tokens.
A token is a piece of a word.
“Carrefour proposal for Q4”
Money. You pay by the token. Your question, and the answer. A long document costs more.
Photo: Declan Sun / Unsplash
Context widows
Everything you give it has to fit
system
tools
your setup
the conversation
…room to work
0context limit →
System instructionsThe vendor's guardrails. Build a bomb? It refuses. And you never see them.
ToolsWhat it may call on. MCP plugs in here.
Your setupRole, house style, templates, skills.
The conversationEvery message, document and result. It stays.
Room to workWhat is left. The answer is written in there.
The box is never empty when you start. And the answer still has to fit in what is left.
04
Confidentiality
Where your words go, and what you can safely paste.
Confidentiality
Three things people mix together
1TransitIs it encrypted while it travels? Yes. Same as your online banking. Not the problem.
2StorageYour conversation is kept. Some tools now remember across sessions. What you write today can come back next week.
3TrainingWill it be used to improve the model? This is the real question.
Confidentiality
A simple traffic light
GreenPublic or generic content. Rewriting, brainstorming, structuring an argument, drafting a generic email.
OrangeInternal content that is not sensitive. Use it, but anonymise first.
RedNamed client data, contracts, negotiated pricing, anything covered by an NDA. Not in an unapproved tool.
05
Hallucinations
Why an answer can be perfectly written, perfectly confident, and completely false.
The most important part of today
Hallucination
Something that sounds completely credible, and is simply false.
Photo: Juli Kosolapova / Unsplash
A hallucination is not a bug.
It is the normal behaviour of the system, applied to a question it does not have the answer to.
It was never trained to say "I do not know"
It was trained to produce a plausible answer.
So when you ask something it does not know, it does not stop. It does what it always does. It produces the most plausible continuation.
Plausible is not the same thing as true.
You ask
What is the total value of the framework agreement we signed with Meridian Foods last year, and which clause covers early termination?
It answers
The framework agreement with Meridian Foods, signed in March 2024, has a total value of €4.2 million over three years.
Early termination is covered by Clause 12.3: either party may terminate with 90 days' written notice, subject to a 15% fee on the remaining committed volume.
The contract is real. Not one of these facts is. It has never seen the document.
Live
Let me show you.
Confident. Well structured. Correct format. Completely invented.
A human who does not know usually hesitates. You hear it in their voice. The model never hesitates. The false answer and the true answer look exactly the same.
Anything about your company. It has no access to it.
Anything recent. It stopped reading at a certain date.
Anything where you push it. Ask for a source and it will often produce one. A source is a plausible continuation too.
Confidentiality
Three habits
1Separate the two types of taskReformulate, structure, summarise what you provided: low risk. Retrieve facts from its memory: high risk. Same tool, different risk.
2Verify anything a client will seeEvery number, every reference. Same check you would run on a junior colleague.
3Give it the information instead of asking for itIf the facts are in the box, it does not have to invent them.
It is not a source of truth. It is a way of working with information you provide.
06
The harness
The model is only the engine. Everything useful is built around it.
A word you will never say out loud
The model is the engine. The harness is the car.
An engine on a table is not transport.
Photo: Garett Mizunaka / Unsplash
Same engine. Different car.
Do not ask what is inside. Ask what it can reach.
07
MCP
One standard plug between the AI and the tools you already use every day.
MCP
A USB-C port between AI and your tools.
Until recently, every connection was custom. One bridge per tool. Expensive, slow, and it broke constantly.
One standard plug. Any tool that supports it connects quickly.
Photo: Mika Baumeister / Unsplash
The AI comes to you.
The tools you already use publish an MCP server. Your assistant plugs in, and works where the work already is.
Hundreds of products ship one today. Your own internal tools can too.
08
Agents
From asking for a task to delegating an objective.
Agents
A model, a harness, tools, and a loop.
It does something, looks at the result, decides what to do next, and continues. Until the objective is reached.
Photo: Homa Appliances / Unsplash
Today, you give an instruction
"Write me a follow-up email for this client."
With an agent, you give an objective
"Follow up with the clients who did not answer this week."
You moved from asking for a task to delegating a goal.
The honest version
This is the area with the biggest gap between the demos and reality.
They work well when the task is well defined and the tools are reliable.
They fail unpredictably when the task is vague.
Mistakes compound. A wrong step two becomes steps three, four and five.
Almost every serious deployment keeps a human at the validation point. The agent prepares. The human approves.
One sentence ties it all together
A model on its own knows nothing about your company, and will invent things confidently.
What we build around it is what makes it useful and safe.
The words
Generative AICreates content, instead of sorting it.
LLMPredicts the next word.
TokenThe pieces it reads, and what you pay for.
Context windowA box. Everything it can see at once.
HallucinationWhat happens when it does not know and answers anyway.
HarnessThe car around the engine.
MCPThe plug that connects it to your tools.
SkillA way of working, packaged once and reused.
AgentAll of it, in a loop, working toward an objective.
Before next session
Test one real question from your own work.
Either
A result that was useful. Tell me why.
Or
A hallucination you caught. Tell me how you spotted it.
We open the next session with your examples, not mine.