Teach on Udemy

Turn what you know into an opportunity and reach millions around the world.

Learn More

Your cart is empty.

Keep shopping

Managing Hybrid Teams of People and AI Agents

Split work between people and AI agents, set how much agents decide alone, design real checks and keep goals fair

Created byCyberProStudy Team

Last updated 11/2026

English

What you'll learn

Decide which tasks go to an AI agent, which stay with people, and how much the agent may decide alone
Design checks people can actually perform, plus measures and alarms that catch drift before customers do
Introduce an agent with fair process: announce it, involve the team and consult employee representatives before a pilot
Own an agent through its whole life: onboarding, probation, updates, demotion and retirement
Keep goals fair when an agent changes the work: re-baseline by case type and make checking work count
Handle an agent's mistake with a six-step escalation path, and prove the agent's value with honest numbers

Course content

10 sections • 43 lectures • 3h 51m total length

Monday morning: your team has changed4:11
Meet Maya, her eight-person team and Kip, the AI agent that arrived on Monday morning. What this course covers, the four principles, and the five role plays.
What an AI agent is, and what it isn't4:34
Three questions that tell an agent apart from a chatbot or workflow automation, and why the moment a system acts creates three new management questions.
Where agents really stand in 20264:22
Where AI agents really stand in 2026: broad use, rare scaling, flat value. Four questions to ask before you repeat any figure.
From checking everything to orchestrating4:12
Maya's first three weeks with Kip: from checking everything to orchestrating. The four principles that make up the manager's new job.
Map your team, including hidden AI use4:05
Build an honest team map, including the AI use nobody told you about. Download: the team map template (D1).
Section Check: Your team just changed

The jagged frontier5:12
The jagged frontier: why you can delegate tasks, never jobs, and why human-plus-AI can be worse than either alone.
Five criteria for every task5:05
Five criteria for every task: repeatability, data quality, error cost and reversibility, room for judgment, emotional weight.
The allocation matrix4:28
Put every task on one page: automate, assist, keep human, or question the task. Download: the allocation matrix (D1).
Autonomy is a dial you set5:02
Five levels of autonomy, three rules for setting the dial, and why the real setting is what happens in practice.
Handoffs that don't drop the ball4:43
Design handoffs between agent and person: when to hand over, what passes with the case, who owns it, and what the customer hears.
When automation went too far5:11
Klarna, Commonwealth Bank, Taco Bell and Cursor: four public reversals and the rule each one teaches.
Section Check: Delegate

Fair process: the one framework5:33
Fair process, the one framework you'll use three times: consistent rules, accurate information, a voice, a way to correct mistakes, an honest explanation.
Announcing an agent (role play 1: Leo)4:41
How to tell the person most affected, first and one to one. Followed by role play 1: Leo.
Telling Leo: the agent takes his work
From affected to involved4:59
Turn affected people into involved ones: the steward role card, and why skills fade when the agent takes the practice.
Psychological safety when the agent arrives5:48
Psychological safety in a hybrid team: a four-step safety loop and the one-to-ones with experimental evidence behind them.
Section Check: Involve

Why a human in the loop isn't enough4:51
Why 'a human in the loop' guarantees almost nothing, and the four conditions that make a check real.
Calibrating trust (Omar's story)5:00
Omar's story: calibrating trust between automation bias and algorithm aversion.
The checkpoint map (D2)4:54
Where to check, how much, and at what cost. The four-stage sampling rule. Download: the checkpoint map (D2).
Measuring the agent4:54
Six measures that tell you whether the agent is doing well, and why conversations handled is not one of them.
Drift, and when to sound the alarm4:43
Drift, test sets and alarm thresholds: when to warn, when to stop, and who can pull the stop.
Section Check: Oversee

Every agent needs a named owner5:19
Every agent needs a named owner in the business, a steward and a technical contact. Download: the agent workbook (D3).
Onboarding an agent5:38
Onboard an agent like a new hire: job description, least-privilege access, the consultation gate, testing and probation. Followed by role play 2. Download: the consultation checklist (D5).
The works council meeting before the pilot
Updates and re-tests4:26
Four ways an agent changes underneath you, and a change routine that keeps each one from becoming an incident.
Demote, pause, retire (D3)4:29
Demote, pause or retire an agent cleanly, and keep every agent in one register. Download: the agent workbook (D3).
Section Check: Own the agent

The selection effect (Amara, Priya)5:25
The selection effect: when the agent takes the easy cases, old averages measure the mix, not the person.
Re-baseline by case type5:30
Rebuild baselines by case type so the numbers measure people again. Three options, from rigorous to simple.
Changing targets fairly5:46
Change targets with fair process: learning goals first, and why a blanket uplift is a trap.
Team goals and visible checking work (D4)5:58
Team goals for shared work, contribution goals for individuals, and checking that counts as work. Download: the fair-goals scorecard (D4).
The performance conversation (role play: Amara)5:16
Prepare and hold Amara's review on like-for-like numbers. Followed by role play 3.
Amara's performance review
Section Check: Fair goals for people

Spotting withdrawal early (role play: Jin)6:04
Spot withdrawal before it shows in the numbers, and open the conversation without judging. Followed by role play 4: Jin.
One-to-one with Jin
Redesign, not replacement5:44
Redesign, not replacement: what moved, what stayed, and the new work the agent created. Redeploy before you cut.
New roles in hybrid teams5:55
Five roles in hybrid teams, from agent steward to complex case specialist, and how to make them real career paths.
Section Check: Roles, engagement and withdrawal

"The agent did it" is no defense6:06
The company answers for what its agents say and do; inside it, a named owner. Victor's case and the Air Canada ruling.
When the agent gets it wrong (role play: Victor)6:14
Stop, scope, notify, fix, tell, log: putting right the harm an agent caused. Followed by role play 5: Victor.
The call with Victor
Bias: patterns to watch6:46
Find patterns that are only wrong in aggregate: a layered sample, outcomes by group, and decisions against clear numbers.
Consult before you deploy7:10
Consulting employee representatives, country by country, and what to bring to the meeting. Download: the consultation checklist (D5).
The EU AI Act calendar and the rest of the world7:01
The EU AI Act calendar for teams using agents, and the main rules outside the EU. Download: the calendar and AI literacy log (D6).
Section Check: Be accountable

Baseline first5:46
Take a baseline before the agent arrives: six to eight numbers, written definitions, measured the same way for people and agent.
Four measure families and the dashboard (D7)6:25
The whole hybrid team on one page: customers, quality and risk, cost, and people. Download: the dashboard workbook (D7).
Reporting upward honestly6:30
Report the agent's value upward honestly: one page, five parts, evidence labels, complete costs, incidents disclosed.
Section Check: Prove the value honestly

Days 1-30: understand5:47
Days 1-30, understand: nothing gets switched on. Six outputs to have by day thirty. Download: the 90-day workbook (D8).
Days 31-60: pilot5:49
Days 31-60, pilot: three gates, a probation period, and a review that can end in no.
Days 61-90: anchor, and the close6:08
Days 61-90, anchor: the monthly rhythm, fair targets, role conversations, a rehearsal, and the record for the next agent.
Section Check: Your 90-day plan
Practice Test: Managing Hybrid Teams of People and AI Agents

Requirements

No technical background or coding needed
Most useful if you lead a team, or will soon, where AI agents are arriving or already at work

Description

This course contains the use of artificial intelligence. All learning content was planned, checked and approved by people.

AI agents are joining teams. They answer customers, draft decisions and hand work to people. That changes the manager's job: someone has to decide which work goes to an agent, how much it may decide alone, who checks it, whose numbers still mean something, and who answers when it gets something wrong. This course is for the team lead or manager who has to make those decisions.

You follow one team as an agent arrives: Maya, who leads a customer service team, and Kip, the agent her company brings in. Along the way you meet her people. Leo fears for his job and becomes the agent's steward. Amara and Priya see their numbers fall for reasons that have nothing to do with them. Jin goes quiet. Omar approves without reading. And Victor, a customer of fifteen years, has his case handled wrong. The course is built on four principles: Delegate, Involve, Oversee and Be accountable.

What you get

43 video lectures, about 3 hours 50 minutes, in 10 sections
A quiz for every section and a 40-question practice test, each answer explained
Five AI role plays: announcing an agent to a worried employee, presenting a pilot to a works council, a performance review, a one-to-one with a colleague who has gone quiet, and a call with an angry customer
Eight downloads: an allocation matrix with a team map, a checkpoint map, an agent workbook (owner card, job description, onboarding record, change log, retirement checklist, agent register, steward role card), a fair-goals scorecard, a consultation checklist, an EU AI Act calendar with an AI-literacy training log, a team dashboard workbook and a 90-day workbook

What the course covers

Your team just changed: what an AI agent is and isn't, where agents really stand in 2026, and mapping your team, including hidden AI use
Delegate: the jagged frontier, five criteria for every task, the allocation matrix, autonomy as a dial you set, handoffs that don't drop the ball, and what happened when companies automated too far
Involve: fair process, announcing an agent, moving people from affected to involved, and psychological safety when the agent arrives
Oversee: why a human in the loop isn't enough, calibrating trust, the checkpoint map, measuring the agent, and drift
Own the agent: a named owner, onboarding with a consultation gate and probation, updates and re-tests, and how to demote, pause or retire an agent
Fair goals for people: the selection effect, re-baselining by case type, changing targets fairly, team goals, making checking work visible, and the performance conversation
Roles and engagement: spotting withdrawal early, redesign instead of replacement, and new roles in hybrid teams
Be accountable: why "the agent did it" is no defense, an escalation path when the agent gets it wrong, bias patterns, consulting employee representatives in Germany, the Netherlands, France, the UK and the US, and the EU AI Act calendar
Prove the value: baseline first, a one-page team dashboard, and reporting upward honestly
Your first 90 days: understand, pilot, anchor

How this course is different

Evidence labels on screen. Every figure is marked as research, a preprint, a survey, a forecast, a company's own report, a vendor's view, law or practitioner judgment, so you know how much weight it can carry.
Vendor-neutral. Nothing to buy: the methods work with whichever agent your company chooses.
The people side, not only the technology: fair targets after an agent changes the work, consultation before a pilot, and roles that lead somewhere.

Up to date

The course was checked in November 2026 against current research and the law as it then stood, including the EU AI Act as amended in 2026. The legal lectures are awareness for managers, not legal advice: check the rules for your country with HR or a lawyer.

Who this course is for:

Team leads and managers whose teams work alongside AI agents, or soon will
Customer service, operations and back-office leaders bringing an agent into existing work
HR partners and change managers who support managers through an AI rollout
Experienced professionals about to lead a team of people and AI agents for the first time

Managing Hybrid Teams of People and AI Agents

What you'll learn

Explore related topics

Course content

Your team just changed5 lectures • 21min

Delegate: split the work task by task6 lectures • 30min

Involve people before the agent arrives5 lectures • 21min

Oversee: checks people can actually do5 lectures • 24min

Own the agent5 lectures • 20min

Fair goals for people6 lectures • 28min

Roles, engagement and withdrawal4 lectures • 18min

Be accountable: errors, escalation, law6 lectures • 33min

Prove the value honestly3 lectures • 19min

Your 90-day plan3 lectures • 18min

Requirements

Description

Who this course is for: