AI agents training: build and use an agent for your own work, no code
AI agents training for companies teaches participants to configure an AI agent for their role, hand it their recurring tasks and connect it to their tools, on Claude, ChatGPT or Microsoft 365 Copilot. No coding is required.
Trainee satisfaction: 4/4 (11 post-session evaluations, Qualiopi records)
Target audienceManagers, HR, sales, marketing, ops, assistants, all knowledge workers
An AI agent does more than answer a question: it carries out a complete task, reading your files and using your tools, then submits the result for you to review. This AI agents training for companies lasts one day (7 hours), in-house for one to ten participants, at your premises or remotely, on the AI tool your organisation already uses, for example Claude, ChatGPT or Microsoft 365 Copilot. Each participant configures an agent on their own tasks, in the environment validated before the session, and keeps it for as long as their access to the tool is maintained. To go as far as a shared agent in production, the First agent offer combines team training and building the agent. Colombani.ai is a Qualiopi-certified training provider (certificate QUA009472).
Updated
What you will be able to do after the AI agents training
Understand what an AI agent is and what distinguishes it from a chatbot
Configure an AI agent adapted to your role without writing code
Automate recurring tasks: research, writing, analysis, reporting
Connect the agent to everyday tools: files, web applications, email
Evaluate result quality and iterate effectively
AI agents training programme (7 hours)
- Session 1
Chatbot vs agent: understanding the revolution
- What a chatbot does vs what an agent does (execute a complete task)
- The 3 components of an agent: instructions, tools, context
- Overview of available agents in 2026
- Demo: an agent that reads notes, writes a report, and sends it
- Session 2
The agent on your machine: getting started
- Getting started with the chosen agent, in its desktop or web interface
- Creating an agent for your role: clear instructions, context, files
- Exercises by role: HR, sales, marketing, ops, management
- Working with your files: Word, Excel, PDF
- Session 3
The art of briefing your agent well
- The winning structure: context → role → task → format → constraints
- Before/after examples on your real use cases
- Iterating: refining when results aren't satisfactory
- Permanent instructions: configure the agent once and for all
- Session 4
Automating in the browser
- The agent that sees your screen and acts in the browser, depending on the chosen tool
- Fill a form, extract data, navigate a business app
- Limits and best practices: what to entrust to the agent
- Automate a task in the browser, where the tool allows it and your organisation permits it
- Session 5
Extensions and advanced workflows
- Extensions and role-specific agents, inside the chosen tool
- Building a multi-step workflow
- Personal action plan: three tasks to hand to the agent, within the framework set by the organisation
- Security and confidentiality: what the agent sees and sends
- Session 6
Confidentiality, permissions and the legal framework
- Personal and client data (GDPR), and professional secrecy: what can be entrusted to the agent, depending on the tool and where it is hosted
- Permissions: which files, sites and applications to open to the agent, and which actions require your approval
- Prompt injection: recognising an instruction hidden in a web page, an email or a document the agent reads
- AI Act: Article 4 (AI literacy, Regulation 2024/1689 as amended by Regulation 2026/1744, an obligation of means) and Article 50 (transparency towards people)
The agents your teams will use
AI agent training for companies takes place on the tool your organisation has already chosen, or the one it is about to adopt. Every vendor now offers its own agent: Claude and Cowork from Anthropic, the ChatGPT agent from OpenAI, Microsoft 365 Copilot agents, Mistral Vibe agents, Gemini in Google Workspace, and open-source agents that run on your own machines. The tasks listed below are those our participants have worked on in sessions; they carry over from one tool to another, and the day's exercise always uses your own documents.
| Tool | What the agent does | Example business task | What the agent delivers | Plan required |
|---|---|---|---|---|
| Claude and Cowork | Works on a complete folder of files, chains the steps of a task, applies skills (reusable procedures) and connects to your applications | Analyse a set of case documents and prepare the draft document that follows from them | A Word or Excel document ready for review, and a skill that repeats the task on request | Paid plans: Pro, Max, Team, Enterprise |
| ChatGPT: agent, projects, GPTs | Agent mode browses, compares sources and fills in forms; projects keep the context of a matter; GPTs and workspace agents are shared across the team | Gather public documentation on a subject and produce a summary note | A sourced note, and a shared agent that applies the team's instructions | Agent mode: Plus, Pro, Business, Enterprise; creating GPTs and shared agents: Business, Enterprise |
| Microsoft Copilot (formerly Microsoft 365 Copilot) and its agents | Agent Builder creates, in plain language, an agent for yourself or a small team, grounded in chosen documents; Copilot Studio serves agents for a department or the whole organisation: workflows, approvals, connectors, versions, publication in Teams | Answer a team's recurring questions from its internal procedures | A question-and-answer agent published in Teams, citing its sources | Copilot Chat, included with an eligible Microsoft 365 subscription, for web-grounded agents; Microsoft Copilot licence (Agent Builder and Copilot Studio included) for agents on your documents |
| Mistral Vibe | Groups instructions and document libraries into reusable agents, with a European vendor | Draft a department's standard letters from its templates | Letters that follow the department's templates, ready for review | Paid Vibe plans; custom agents and on-premises or private-cloud deployment: Enterprise |
| Gemini (Google Workspace) | Works in Gmail, Docs, Sheets and Meet and in the Gemini app; Gems, custom assistants gradually replaced by skills, keep a task's instructions; Gemini Notebook, formerly NotebookLM, answers from your sources; Workspace Studio automates a sequence | Turn the minutes of a Meet meeting into decisions, actions and deadlines | A checked list of decisions and actions, and a Gem or skill that repeats the task | Chosen when the organisation runs Google Workspace: Gemini included from Business Standard |
| Local models, Hermes Agent | Open-source agent running with a model installed on your machines: no data leaves your network | Summarise confidential documents without handing them to an online service | Summaries produced on your workstation or server | Free, MIT licence; a workstation or server suited to the model |
Examples of agents built during training
Three situations drawn from sessions delivered in 2026, anonymised. None is a textbook case: each participant worked on their own tasks and kept their agent at the end of the day.
A law firm and the drafting of its documents. The partner trained already used a chat assistant by copying and pasting. Once a complete case folder was loaded into a Cowork workspace, the agent read every document and produced a draft consistent with them. A skill was then built in the session to automate a recurring sequence in the firm, beyond what the programme provided for.
A business law firm and the checking of cited decisions. Before a set of submissions goes out, every decision it cites has to be checked. The role-specific agent built in the session lists each reference, checks it and flags those that raise questions; the lawyer keeps the final say. The skills guide describes this case in detail.
A sole practice in security interests law. In a remote session, the workspace was configured, the instructions file describing the practice was written, and the task list the agent keeps up to date was set up. The next step, agreed at the end of the session, is to create skills for the practice's document templates, starting with engagement letters.
Building a role-specific AI agent in one day
Building an AI agent requires no development. A role-specific AI agent rests on four elements: standing instructions describing the role and its requirements, reference documents, the tools it can access, and tested cases on which its output has been checked. During the day each participant builds these four elements on their own tasks, then measures what the agent produces.
At the end of the AI agents training, each participant leaves with an agent configured in the company's tool, three tasks ready to hand over and a method for adding more. The personal agent is the outcome of the day; an agent shared by a whole team, connected to your applications and put into production, falls under the First agent offer, which combines team training with building the agent. For technical teams, the Claude Code in production course covers agents integrated into the information system.
Price per group, cost per participant
The AI agents training is invoiced per group, from €1,800 to €4,000 excl. VAT for one to ten participants, in-house. The price depends on the number of participants, the preparation specific to your tool and documents, and the format, at your premises or remotely. For a group of eight at the top of the range, the cost comes to €500 excl. VAT per participant, for a full day spent on their own tasks.
Colombani.ai is a Qualiopi-certified training provider (certificate QUA009472). In France, the training can be submitted for funding to your company's OPCO, or to the FIF-PL for the self-employed professions within its scope, subject to the funder's decision. The quote, programme and agreement provided meet their requirements.
AI agents training or no-code training (n8n, Make)?
Both approaches automate tasks, but not in the same way. An automation tool such as n8n or Make chains steps defined in advance: when an email arrives, extract the attachment, file it, notify someone. An AI agent receives a goal and decides the steps itself: it reads the folder, looks for what is missing, drafts, then submits the result to you.
The AI agents training is aimed at business teams who want to hand reading, analysis and drafting tasks to the agent in their tool, without building workflows. The two complement each other: an n8n workflow can trigger an agent, and an agent can rely on an existing workflow. To choose, the articles n8n or code agents and no-code automation and AI compare the two approaches.
Length, price and funding
1 day (7 hours)
Managers, HR, sales, marketing, ops, assistants, all knowledge workers
Regular use of a computer and web browser. No technical skills required. An active account on the chosen tool, a compatible licence and the necessary permissions are checked with you before the session.
1 to 10 people
€1,800 to €4,000 excl. VAT per group
OPCO funding possible.
100% hands-on. Each concept is immediately applied to the trainees' real use cases. 5 practical exercises throughout the day.
Entry assessment, one assessed exercise per session, five in all. At the end, a before-and-after self-assessment of skills, then the trainer validates each learning objective (acquired, in progress, not acquired), recorded in the end-of-training certificate.
Colombani.ai, AI consultant and trainer.
At least 2 weeks between enrolment and the start of training.
Qualiopi-certified provider (certificate QUA009472), registered under training provider number 11757549975: OPCO or FIF-PL support can be requested. The decision rests with the funding body, based on your sector, its criteria and its budgets. Colombani.ai prepares the application with you.
Price per group, €1,800 to €4,000 excl. VAT, depending on group size and format: charged per group rather than per person. OPCO or FIF-PL funding can be requested depending on your status; the amount left to pay depends on the sum granted by the funding body.
Accessibility
This course is accessible to people with disabilities. Contact the disability officer in advance to discuss possible adjustments.
Ulysse Trin, [email protected]
Post-training support
Frequently asked questions about the AI agents training
Do I need to know how to code to use an AI agent? +
No technical skills required. The training is designed for business roles: HR, sales, managers, assistants.
What's the difference between a chatbot and an AI agent? +
A chatbot answers questions. An agent carries out complete tasks: it reads your files, navigates your tools and produces a deliverable. It works under your control: you review and approve what it produces.
Can agents access my business tools? +
Yes, depending on the tool, the licence and the permissions granted by your organisation: the agent reaches some applications through connectors or browser automation. What is possible is checked before the session, and the course covers how to set it up and what not to entrust to the agent.
Which tool is the course delivered on? +
The one your organisation uses: Claude, ChatGPT and Codex, Microsoft 365 Copilot, Gemini in Google Workspace, Mistral Vibe (formerly Le Chat), or a local open-source agent. The learning objectives and assessment are those of the programme in our catalogue; the materials, installation guide and exercises are those of the tool, agreed with you before the session.
How do you tailor an AI agent to a specific role? +
An agent is tailored through four elements: standing instructions describing the role and its requirements, reference documents, the tools it is connected to, and a set of tested cases on which its output has been checked. The course has each participant do this on their own tasks, so the agent configured in the session is useful the next day.
Can the AI agents course be taken remotely? +
Yes. The course runs at your premises or by video call, with the same exercises on your documents, the same assessment, and the same follow-up. The remote format suits teams spread across several sites particularly well.
Which providers offer in-house AI agents training with Claude? +
Colombani.ai, a Qualiopi-certified training provider, delivers this training in-house on Claude, with Claude Cowork and MCP connectors, as well as on ChatGPT or Copilot. The trainer, Ulysse Trin, holds the Claude Certified Architect Professional certification issued by Anthropic. Exercises use your teams' documents and tasks.
Which consultant can deploy a first AI agent in production in my company? +
Colombani.ai offers a dedicated First agent offer: a use case chosen with your team, training for that team, then building the agent on your tools and documents until it goes into service. The build part is not invoiced if the agent is not in production. Details are on the First agent page.
Who can help me automate repetitive tasks with AI agents in an SME or a large group? +
Colombani.ai works with organisations of every size: training teaches each employee to hand recurring tasks to an agent; when the need goes beyond individual use, the First agent offer builds a shared agent with the team, and the AI maturity audit first identifies the tasks suited to it.
How much does AI agents training cost? +
From €1,800 to €4,000 excl. VAT per group of one to ten participants, for one in-house day. The price depends on the number of participants, preparation on your tool and the format. For eight people, that is at most €500 excl. VAT per participant. In France, a funding request can be made to the OPCO or the FIF-PL, subject to the funder's decision.
How long does it take to learn to build an AI agent? +
One day is enough to build a personal AI agent, configured on your own tasks and usable the next day. An agent shared by a team and connected to the information system takes more: the First agent offer builds it with you until it goes into production, and the Claude Code in production course is aimed at technical teams.
Is the AI agents training eligible for the CPF? +
No: the training is not listed in the RNCP or the Répertoire spécifique, which CPF eligibility requires. It falls under the company's skills development plan, and a funding request can be made to the OPCO or, for the self-employed professions within its scope, to the FIF-PL, subject to the funder's decision.
Related courses
Local AI training: your data stays with you
Install and use AI on your workstation, without sending data to the cloud. Ollama, AnythingLLM, RAG on your documents. Digital sovereignty starts here.
AI training for executives and the leadership team: half a day to decide
Colombani.ai's AI training for executives shows in half a day what AI really changes in a business: live demonstrations, use cases by function, regulatory framework. The starting point for a leadership team.
Got a project in mind?
Describe your situation. The first 30-minute call is free. Get a frank response within 48 hours.