.webp)
TutorialHow RAG Transforms Your AI into an Expert on Your Business
RAG and LLM: definition, how it works, choosing the model. The guide to connecting an LLM to your internal documents and enterprise data.
Training & enablement
AI training for your teams,entirely bespoke.
We come from both management and engineering, and we train on business cases you can put a number on. Every session is written after a scoping call, at the actual level of the people who will be in the room. On site in Paris or remote.
By Jean-Christophe BudinUpdated on
Two kinds of expertise, and a gain at the end
Business cases with the gain attached
A generic AI course explains what an LLM is. Fine. But once the room empties, the question left standing is always the same: what does this change here, for our business, with our data and our constraints?
We bring expertise on both sides, management and engineering. We have run and managed a business, and we also build the systems. That combination is rare, and it is what helps most: most of the blockers people describe to us are not technical, they are two teams using different words for the same problem.
Our position is that AI is only worth it if it returns something you can measure. We work on identified business cases with the expected gain next to them: hours recovered, errors avoided, lead times cut. And when the numbers say plain automation would do better for less, we say that too.
Fully bespoke
Content and examples get written after the scoping call. Two clients in the same industry do not get the same session.
On site in Paris or remote
At your offices in Paris and the Île-de-France region. Remote when your teams sit across several sites.
Management and engineering
An architecture constraint becomes a margin question. A business need becomes a technical choice. We hold both ends, in front of the same audience.
Content
What we cover
To be assembled depending on the audience and what you want to get out of it.
The concepts, minus the jargon
LLMs, embeddings, RAG, agents, fine-tuning, context windows, hallucination. Those notions laid out clearly, without overselling them.
Under the hood, live
What happens between the question and the answer. Why a model gets it wrong. What changes when you swap a source or a prompt. Hands on the keyboard, not commentary over diagrams.
What it does well, and what it fumbles
The uses where the technology delivers today, the ones where it disappoints, and the ones where a deterministic script would do better for less. With the costs and risks attached.
Costing the value before you commit
How to estimate the gain of a use case: what you can measure, what you will never measure, and the threshold above which it is worth doing at all.
On your own ground
Every notion is replayed on your cases. A quality manager and a lawyer do not ask the same question about hallucination, and do not need the same answer.
Demonstrations on your own cases
We show what AI produces on your documents and your use cases. Watching an idea run on your own data beats one more slide.
Generative AI, agents and automation
Training on generative AI, AI agents and automation. What can be handed to an agent, where to keep a human in the loop, how to measure the gain without kidding yourself.
AI inside your development process
Coding assistants, automated review, test generation: where your teams actually save time, where they lose it, and what it changes for code quality and review.
Audiences
You do not talk about AI the same way to everyone
Same subject, but neither the same level of detail nor the same examples.
Executive committee and leadership
AI training for the C-suite: where to invest and for what return, what risks you are taking, how fast to move, what sovereignty actually changes. Little technical detail. The subject is the trade-offs.
IT leadership and engineering teams
AI enablement for the IT department: architectures, security, inference costs, data governance. We also cover AI inside your development process, and the industrialisation traps you only see after shipping a first POC.
Managers and business teams
What AI changes in a working day: concrete uses on the tools they already have, limits worth knowing, and the moments when the answer deserves suspicion.
Independence
We are not AI evangelists
In plenty of cases, the right answer to a process that is stuck is not AI. It is deterministic automation, simpler and more reliable, or a redesign of the process itself. We own that: we only put AI where it genuinely adds something.
Ask This Guy also builds its own AI platform. That is a fair objection, so let us deal with it up front: our training is not a disguised product demo. We present and compare the other platforms on the market, proprietary and open models, American and European offerings.
Your teams should leave able to choose. When neither AI nor Ask This Guy is the right answer to your need, we say so during the session.
Partnership
Alongside your AI consultants
Already working with a consulting firm or a systems integrator on your AI strategy? We are not there to take their place.
We step in as a complement, on the technical and educational side: explaining how the building blocks really work, discussing the assumptions behind a scoping exercise, putting a demonstration behind a recommendation. Usually we get called when something has to be decided.
Our offer for integrators and AI consultantsOur services
Beyond training
Plenty of clients call back after the training. Usually for one of these reasons.
A demonstration in your context
A demonstration built on your data. Useful when an internal debate is going in circles and people finally need something to look at.
Request a demonstrationAI scoping and roadmap
Use cases ranked by value, an AI roadmap and a governance framework: data access rights, answer traceability, GDPR and the AI Act.
Our CIO offeringCo-development and forward deployed engineers
Our engineers work at your place, with your tools and your data, to take a POC all the way to production. The code and the architecture choices stay yours.
From POC to productionThe Ask This Guy platform
Document RAG, agents, connectors and EU hosting, when you would rather start from a ready-made foundation than build everything.
Explore the platformHow it runs
What a session looks like
- 1
Scoping
A conversation to pin down your subject, find out who will be in the room and at what level, and what those people should be able to do afterwards.
- 2
Design
We write the running order, take the examples from your industry and prepare the demonstrations on your cases.
- 3
Delivery
On site in Paris or remote. A short format for an executive committee, longer and more hands-on for a technical team.
- 4
Follow-up
The session materials, and answers to the questions left hanging. If you want to go further, that is when we talk about it.
Resources
Go further
Selected articles and videos to better understand this solution and see how it works in practice.
Articles
Tutorial
GuideEnterprise AI agents: use cases and real costs
Enterprise AI agents: the definition, what an agent can really do, the method that gets it into production, and what it actually costs.
GuideGetting AI automation right: use as little AI as possible
What is AI automation? Benefits, examples, tasks worth automating, and key steps to do it right while using as little AI as possible in production.
AILLM On-Premises for Your Enterprise: What Can You Do with €60k?
Agents, RAG, internal assistants: how to choose between Make and Buy for your AI project. Figures, use cases, and cost analysis to help you decide.
4 min
GuideKnowledge management: why your initiatives fail, and what AI changes
Why knowledge management initiatives fail, what AI really changes about managing what a company knows, and what it will never change.
ConceptsOpen-Source in AI: The Big Misunderstanding
Open-source vs open-weight: why LLMs can't be truly open-source like Linux, and what you actually get when a model releases its weights.
Frequently asked questions
Do you offer a catalogue of AI courses?
No, bespoke only. Every session is written after a scoping call, from your own subjects and the actual level of the people attending. A standard module would produce a session your teams have already seen elsewhere.
How do you measure the return on an AI use case?
By starting from the existing process rather than from the technology. We look at how long it takes today, where it produces errors and what those errors cost, then estimate what AI takes out of that. Some gains cannot be measured, and we say so too: three costed cases beat a list of twenty ideas.
Do you always recommend using AI?
No. On many processes, deterministic automation or simply redesigning the process gives a better result at lower cost and lower risk. We would rather say so during the training than let a team launch an AI project that does not justify itself.
Where do your enterprise AI training sessions take place?
On site at your offices in Paris and the Île-de-France region, or remotely for teams spread across several locations. On-site works better for hands-on workshops, remote suits awareness and enablement sessions well.
Who are these training sessions for?
Executive committees, IT leadership and engineering teams, and managers and business teams. Wearing both a business and an IT leadership hat mostly helps us bridge the two inside the same organisation.
Do you only train on your own platform?
No. We present and compare the other platforms on the market, proprietary models as well as open ones, American offerings as well as European ones. The goal is for your teams to be able to choose knowingly, not to leave convinced they should buy Ask This Guy.
Can you work alongside our AI consultants?
Yes, often. We step in alongside a consulting firm or an integrator already in place, on the technical and educational side: explaining how the building blocks really work, discussing the assumptions behind a scoping exercise and putting a demonstration behind a recommendation.
Can we see AI running on our own data?
Yes, and that is what we recommend. We prepare demonstrations on your documents and your use cases, so the discussion is about observable results rather than general principles.
Tell us what you need to get across
Thirty minutes is usually enough to tell whether we are the right people for your subject.