Generative AI training is only as good as how differently participants do things the following Monday. Our entire teaching approach stems from this idea: you don't learn AI by listening to it, you learn it by practicing it on your own work, using your company's tools. Here's how we structure and facilitate each session, from the Prompt Party® one-hour session at the Studio over six days.
The NextStart method, in brief
- The principle
- Guided practice from start to finish. No lecture, no demonstration.
- The group
- A maximum of twelve participants per workshop, each on their own computer, working on their real-life cases.
- The tool
- Yours: Copilot, ChatGPT, Claude, Gemini, Mistral, or your secure internal assistant. We provide training within your environment, without any third-party tools.
- The progression
- Party (1 hour or ½ day), Challenge (1 day), Hackathon (2 days), Studio (6 days).
- Deliverables
- Training booklet, library of prompts and agents by profession, access to the NextStart library.
- Measurement
- Immediate satisfaction, then adoption at day 30 and day 90 upon request.
- The results
- More than 4,850 professionals trained in around thirty organizations, average rating of 9,4 out of 10.
Four biases
Practice before talk
Each concept is followed by an exercise. Participants spend most of the session with their hands on the tool, guided by the instructor, rather than looking at slides.
Your real-life cases, not generic examples
The exercises are built from time-consuming tasks collected beforehand from your teams. We work on a real report, a real call for tenders, a real data table.
Your tool, in your environment
We lead workshops on the AI deployed in your environment, including its settings and limitations. What participants learn in the classroom works identically at their workstations the next day.
Safety first
Each session begins with what can and cannot be entrusted to the tool: consumer accounts and professional offers, use of data for training, classification of your information.
A recognized skills framework: Anthropic's AI Fluency
To describe what each participant needs to know when interacting with AI, we rely on Anthropic's AI Fluency framework. It's tool-independent: it applies to Copilot, ChatGPT, Claude, Gemini, and Mistral. It identifies four skills, the "4Ds," which our workshops practice using real-world scenarios.
Delegation
Decide what belongs to AI, what remains for humans, and what is done by two people.
Description
Clearly state the task, the context, and the expected result.
Discernment
Evaluate the accuracy, quality, and biases of what the AI produces.
diligence
To take responsibility for usage, transparently and securely.
Flavien Besset holds the certificate Teaching the AI Fluency Framework delivered by Anthropic, which prepares to teach and assess these four skills in training. View his certifications
Five steps, from gathering needs to measuring usage
- Collection needs to be identified from a subject matter expert, or through an upstream survey of participants: current level, time-consuming tasks, expectations.
- Proposition a process adapted to your context, your AI tool and your security constraints, reviewed with you until validation.
- Animation either in person at your premises or remotely, in French or English, by a trainer who uses AI in business on a daily basis.
- Publications : training booklet, library of prompts and agents by profession, access to the NextStart library.
- Measured from adoption to J+30 and J+90 on request: real uses, time saved, use cases to industrialize, to decide on the next steps.
To showcase everything that AI can do
When we observe actual practices within an organization, two skills dominate: writing and summarizing. Our workshops provide practice for the other six, the ones that truly transform work: researching with verifiable sources, analyzing data, brainstorming methodically, receiving coaching before a decision or interview, designing, and coding. This is why a Prompt Party® is much more than a prompting course.
The two already widespread uses are shown in dotted lines; the six additional skills that our workshops practice are shown in color.
A progression by formats
The same formats exist for all four ranges: the prompt, the AI agents, Cowork and the codeWe start with a short format, and increase the intensity once the practices are established.
| Format | Duration | What we do there |
|---|---|---|
| Party | 1 hour or half a day | The breakthrough and initial reflexes in one hour; several business cases worked through in half a day. Deployable in waves to cover a team, a site, or an entire organization. |
| Challenge | 1 day | A day spent on real-life missions, with results presented at the end of the session. |
| Hackathon | 2 days | Teams build agents or delegation flows on their own processes. |
| Studio | 6 days | Support all the way to agents and processes that run at your location. |
Scaling up with internal relays
Training several hundred employees cannot be done solely with external trainers. The process AI Ambassadors forms a collective of referents who then themselves lead Prompt Party within their respective entities, they were provided with a ready-to-use facilitation kit and co-facilitated the initial sessions. In one organization, a group of 10 ambassadors co-facilitated 84 workshops across five waves. AI Ambassadors Kit is free to download.
What this looks like on the ground
All Detailed customer case studies are online. For administrative information (business registration, documents, funding), see the page Training organizationAnd regarding the prices, How much does AI training cost?
An AI training project?Written proposal within 24 working hours, without obligation, after a 30-minute framing exchange.
Let’s Talk About Your Project