AI Agents

An agent is not a better chat: it is an assistant you delegate a task to, from start to finish. This training shows where that pays off, where it does not, and how to build the first ones safely.

Request a Proposal

Who This Training Is For

• Teams already using generative AI who want to move from questions to delegated tasks
• Heads of operations, customer service and administrative processes
• Product and technology teams assessing where to automate
• Organizations that want to understand what an agent can and cannot do

What Your Team Will Be Able to Do

• Tell a one-off request apart from a task worth delegating to an agent
• Design an agent from a real process in the organization
• Connect an agent to the team’s information sources and tools
• Define the points where a person must validate
• Measure whether the agent is actually saving time, with numbers

Programme

1. What changes with agents — from conversation to execution, and why that forces you to think in processes
2. Anatomy of an agent — instructions, memory, tools and limits
3. From process to agent — pick a case, map the steps, decide what stays with the person
4. Guided build — each team builds an agent for its own process
5. Control and trust — human validation, an audit trail, and what to do when it fails
6. Governance — what data it may see, what decisions it may not take, who is accountable

Learning Objectives

• Distinguish traditional automation, AI-assisted automation and agent-based automation, and identify when each applies
• Create and configure custom AI agents (role, goal, instructions, rules and tone) in ChatGPT, Microsoft Copilot and Claude
• Understand what data and information make an AI agent specialised and useful, and what should be avoided
• Automate recurring research and analysis using AI-augmented search tools
• Build a customer support chatbot or assistant on a no-code automation platform (Zapier)
• Configure task automations and agents in Make, connecting different applications and decision steps
• Apply what was learned by building your own assistant or automation, in a final practical challenge

Format

• In person or online
• Adaptable duration: 4h, 8h or 16h depending on depth
• Every participant leaves with an agent built for their own work
• Cases adapted to the organization’s real processes

What Makes It Different

Most agent demos use examples that look like nobody’s actual job.

Here we start from a process inside the organization — one of those that eats hours every week — and build on top of it. What leaves the room is ready for Monday.

Are there tasks in your team that repeat every single week?

Request a tailored proposal.