AI Agents Training

From Chat to AI Agents:
What Should You Use, When?

AI is no longer just about asking a chatbot a question. New tools can work alongside you, complete multi-step tasks and turn repeated work into reusable agents. This practical, non-technical session helps teams understand the difference and choose the right approach for the job.

Discuss Team Training Explore AI Business Training
Practical and non-technical
Vendor-neutral principles
Built around real business tasks
Infographic showing four stages of leadership delegation: Ask, Collaborate, Delegate, and Operationalise, with illustrated people and icons.
Start Here

Not Everything Needs an Agent

Chat is not an outdated version of an agent. These are different ways of allocating responsibility between a person and the AI, and more autonomy is not automatically better.

More autonomy can help when work is multi-step, coordinated or repeatable.

More autonomy can also add complexity, risk and unnecessary effort.

The right choice depends on the work, not which technology sounds newest.

The Core Model

Four Ways to Work with AI

These are four different ways of working, not four levels of maturity. You do not need to progress through them in order, and most people will move between them from task to task.

Ask

"Help me"

Chat

A conversational way of working where the person remains the main driver of the work.

Best for

  • Answers and explanations
  • Ideas and summaries
  • Short drafts
Collaborate

"Work with me"

Co-working AI

Human and AI work together on a substantial piece of work, with the person shaping and refining the result.

Best for

  • Substantial documents
  • Analysis and planning
  • Work needing iteration
Delegate

"Do this for me"

Agentic AI

AI is given an outcome and can determine and perform several steps, use tools, gather information and adapt with less continual prompting.

Best for

  • Multi-step work
  • Gathering information
  • Completing a defined outcome
Operationalise

"Do this whenever it needs doing"

AI Agent

A configured, reusable AI capability designed for a role, task or process, combining instructions, knowledge, tools, permissions, actions and governance.

Best for

  • Repeatable workflows
  • Recurring processes
  • Shared team capabilities

Choose the lowest level of autonomy that reliably gets the job done. A simple task does not become better because an agent performs it.

Decision Path

What Should You Actually Use?

Four things usually determine the right choice: how often the work happens, how many steps are involved, how much judgement is needed and what happens if AI gets it wrong.

Need an answer, idea, explanation or short draft?

Chat

Need to build and refine substantial work together?

Co-working AI

Need AI to determine and perform several steps?

Agentic AI

Does the same workflow repeat regularly?

Consider an AI Agent

If the task is high-risk, ambiguous or difficult to reverse:

  • Reduce autonomy
  • Restrict access
  • Add approvals
  • Keep a human accountable
Session Content

What the Training Covers

A practical, non-technical session focused on business judgement. This is not a coding course, and no technical background is needed.

  • Understand Chat, co-working, agentic AI and AI Agents
  • Choose the right approach for different types of work
  • Identify useful and poor Agent opportunities
  • Decide how much autonomy and human oversight makes sense
  • Apply the principles to ChatGPT, Microsoft Copilot, Claude or Gemini

The session is vendor-neutral first. We teach the underlying work patterns before translating them into the tools your team uses, so the principles stay useful even as products change.

Practical Non-technical Focused on business judgement
Training Formats

AI Agents Training to Suit Your Team

This topic can be delivered as a focused standalone session, or incorporated into a broader AI training programme. Sessions run online or in person, whichever suits your team.

Focused 2-Hour Session

A focused specialist session covering the four ways to work with AI and how to choose between them.

Part of a Half-Day Workshop

Included as a module within a half-day AI workshop, alongside practical application and hands-on exercises.

Part of a Full-Day Programme

Built into the wider full-day AI training programme for teams who want a broader foundation.

Formats and pricing are set out in full on our AI Business Training page, so everything stays in one place and up to date.

View Training Formats and Pricing
Applied to Real Work

The Same Work Can Need Different AI

The right approach depends on the version of the task, not just the job category. These examples are vendor-neutral and illustrate the pattern.

Chat

One important follow-up email

Why: The person already has the context and final judgement matters.

Co-working AI

Developing a substantial strategy, proposal or campaign

Why: The value comes from iteration, challenge and refinement with the person remaining closely involved.

Agentic AI

Preparing tomorrow's important client briefing using emails, notes, documents and external research

Why: The outcome is clear, but gathering and synthesising the information involves several steps.

AI Agent

Preparing the same type of client briefing for every account director before every strategic meeting

Why: The task has become a stable, repeatable process.

Autonomy and Risk

More Autonomy Is Not Always Better

Giving an AI system more responsibility means it can access, decide and act more. These are materially different levels of responsibility:

Read Analyse Recommend Prepare an action Act with approval Act without approval
Permissions Human oversight Approvals Accountability Safe-use guardrails

Some work should stay human-led. And if a task follows fixed rules every time, conventional workflow automation may be a better answer than an AI Agent.

Explore Using AI Safely at Work training
Who It's For

Useful for Teams, Managers and Leaders

The training can be adapted for different levels of responsibility, and delivered to mixed groups where that suits you better.

Business Teams

Understand the different ways AI can support work and choose the right approach for everyday tasks.

Managers

Identify work that may be suitable for delegation, repeatable agents or continued human control.

Leaders

Build a clearer view of where agentic AI may add value, where guardrails are needed and where autonomy should remain limited.

FAQs

Common Questions

Practical answers for teams thinking about AI agents training.

Choose the Right Level of AI for the Work

The goal is not to use the most advanced AI option. It is to use the approach that gets the job done with the right balance of capability, judgement and control.

Discuss Team Training Explore All AI Training