Skip to main content
Skills course

Agentic AI at Work: Designing, Running and Supervising AI Agents

Learn to put AI agents to work and stay accountable for them: choose the right work, design an agent's job, set its permissions, and test, supervise and measure what it does.

  • Format Online, self-paced
  • Scope 6 modules and a capstone
  • Duration About 8 h
  • Cost Quote on request

About this course

This course teaches how to design, run and supervise AI agents: systems that act on your behalf rather than only answer. It starts with what an agent is and what it is good and bad at today, then covers the three tests for choosing its work, how to write its goal, scope and stop rules, how to give it the access it needs and nothing more, and how to test it, read its logs, measure it and pause it when something goes wrong. A final module covers your organization’s policies, telling colleagues and customers when an agent is acting, and the EU AI Act’s accountability rules. The capstone is one supervised agent for a real process, with its job description, permissions, test plan, monitoring plan and a handover note.

Agents are arriving in the tools people already use: they triage tickets, book meetings, chase invoices and file documents on their own. The organizations that gain from them are the ones where someone can say which work an agent may do, under whose identity, with what access, and who checks the result. Without that, agents get deployed by whoever is quickest, with too much access and nobody watching. Learning the discipline gives the organization agents that save time and stay within bounds, and gives every process owner a way to use them without waiting for engineering.

  • Practice on real tasks and an applied assignment
  • Graded final assessment

A certificate that lists the skills is issued when the final assessment is passed and the applied assignment is submitted.

Content

  1. Module 1

    What an Agent Is, and What It Is Not

  2. Module 2

    Choosing the Right Work for an Agent

  3. Module 3

    Designing an Agent's Job

  4. Module 4

    Permissions, Data and Security

  5. Module 5

    Running and Supervising Agents

  6. Module 6

    Agents in the Organization

  7. Module 7

    Capstone: Design and Supervise an Agent

Apply it in a workshop

A half-day with an expert, live online or in person, that puts this course to work on your team's own cases. The course is the pre-work; the workshop is where it gets applied.

Who it is for

  • Operations, product, HR, finance and support staff who own a process
  • Team leads
  • Career-changers into operations and automation roles

Fee and registration

Quote on request

Price Tell us which Skills courses and how many learners, and we send you a quote. Larger orders get volume pricing, and there is no minimum purchase.
Quoted on request
What a license covers Any license can be used on any Skills course: you choose the course when you assign it to a named learner.
One learner on one Skills course
What you need Skills course licenses sit on top of user licenses: a learner has to be a user on the platform.
A TechClass subscription
Access A license is used when it is assigned, whether or not the learner finishes. Unused licenses stay valid for 12 months from purchase.
12 months from assignment

How to get started

  1. Request a quote

    Tell us the Skills courses and the number of learners. We come back with a quote for the licenses and can walk you through the courses in a demo.

  2. Add your people

    Every learner needs a user license on TechClass. Skills course licenses are added on top.

  3. Assign a license to each learner

    Your admin assigns a license from the pool to each named learner on this course. The learner then has 12 months to complete it, at their own pace.

Education organizations and continuous learning departments: a plan of their own, paid per completion and with learner support, is in preparation. Contact us to discuss your program.

Request a quote
Questions, answered

Agentic AI at Work: Designing, Running and Supervising AI Agents FAQ

Per learner, per course. You buy Skills course licenses as a pool, on top of your user licenses, and your admin assigns one to a named learner on whichever Skills course they need. A license is used when it is assigned, whether or not the learner finishes.

Skills course licenses are quoted on request. Tell us which courses and how many learners you have in mind, and we send you a quote. Larger orders get volume pricing, and there is no minimum purchase.

Yes, with one license for each course. Licenses come from the same pool, so any license can be assigned to any Skills course, and all the licenses in one order count towards the volume price.

About 8 hours of guided learning, in modules, at the learner's own pace. Access lasts 12 months from the day the license is assigned.

Yes. A certificate that lists the skills is issued when the final assessment is passed and the applied assignment has been submitted.

A piece of real work that puts the skills to use. An employee does it on their own job and team; a learner without a suitable workplace does it on a provided case company. The learner checks it against a rubric, and a manager can review it.

No. Every Skills course stands on its own and restates what it relies on, so a learner can start with it directly.

The course is the same for companies and education organizations. A plan for continuous learning departments, paid per completion and with learner support, is in preparation.

English, at this stage.

Artificial Intelligence

Interested in Agentic AI at Work: Designing, Running and Supervising AI Agents?

Learn to put AI agents to work and stay accountable for them: choose the right work, design an agent's job, set its permissions, and test, supervise and measure what it does.

Full table of contents

Agentic AI at Work: Designing, Running and Supervising AI Agents

7 modules · 24 sections

Module 1 What an Agent Is, and What It Is Not

  1. 1.1 From assistants that answer to agents that act
  2. 1.2 Tools, memory, goals and loops: how an agent works
  3. 1.3 Where agents already sit: email, calendars, tickets, files, browsers
  4. 1.4 What agents are good and bad at today

Module 2 Choosing the Right Work for an Agent

  1. 2.1 Repeatable, rule-bound, reversible: the three tests
  2. 2.2 Mapping a process step by step
  3. 2.3 Where a person must stay in the loop
  4. 2.4 Estimating the value and the risk

Module 3 Designing an Agent's Job

  1. 3.1 Writing the goal, the scope and the stop rules
  2. 3.2 Giving it the tools and data it needs, and nothing more
  3. 3.3 Instructions, examples and guardrails
  4. 3.4 Handoffs: when the agent asks a person

Module 4 Permissions, Data and Security

  1. 4.1 Identity: on whose behalf an agent acts
  2. 4.2 Least privilege for agents: access, approvals and spending limits
  3. 4.3 Confidential and personal data in an agent's hands
  4. 4.4 Prompt injection and other ways agents get tricked

Module 5 Running and Supervising Agents

  1. 5.1 Testing before go-live: dry runs and sandboxes
  2. 5.2 Logs, audit trails and reviewing what the agent did
  3. 5.3 Measuring accuracy, cost and time saved
  4. 5.4 When it goes wrong: pausing, rolling back and reporting

Module 6 Agents in the Organization

  1. 6.1 Policies: who may deploy an agent, and for what
  2. 6.2 Telling colleagues and customers when an agent is acting
  3. 6.3 The EU AI Act and accountability for automated actions
  4. 6.4 Keeping agents maintained as tools and models change

Module 7 Capstone: Design and Supervise an Agent