
The lecture starts with real work: you take one real working week, write down the tasks in it, and place each task on a map that tells you what to hand over, what to keep, and how to check what comes back.
This course is built from six programmes: everyday assistant basics, prompt architecture and context, applied use cases across a whole function, a second assistant for documents and analysis, a third assistant inside a document suite, and agents.
The map of the course
A simple dictionary for reading lectures from other fields as your own work
Three questions to ask after every lecture
A short self-diagnostic that shows which sections to watch first
Download the Give or Keep Map and fill in one line after each lecture.
What these models are: what they can do and where they stop
A short history, and why the current generation is different
What alternatives exist and how to choose between them
Formulating a request properly, and where all of this is heading
Organising tasks and plans with an assistant
Everyday problems it genuinely solves
Generating options and supporting a decision
Optimising your own repeated processes with templates
Business writing: letters, reports, instructions, templates
Cutting the routine part of a working day
Analysing information and files you upload
Building a personal assistant on your own documents, tuned to your role
Generating ideas: concepts, names, wording
Content for channels and projects
Rewriting and adapting text to a required style
Producing images alongside text
Finding and structuring information
Summaries, lists and analytical material
Learning a new subject through dialogue rather than reading
Practising conversation, and improving your own writing
Integrating the assistant into everyday tools and automation
Building complex prompts that return precise answers
Non-standard tasks, from wording to strategy
Services built on top of these models
Role, context, task, examples, constraints, format, validation
Four frameworks — RISEN, CRISPE, CREATE, RTF — and when each fits
Why role and context account for most of the answer quality
Few-shot examples, and making the model check itself before answering
Chain-of-thought and what it adds on judgement tasks
Self-consistency: several answers, and picking the reliable one
Tree-of-thoughts for multi-criteria assessment
Reasoning plus acting, and combining techniques in one prompt
The shift from prompt engineering to context engineering
Building an assistant loaded with your own documents and policies
Custom instructions: set up once, works every time
Retrieval explained without an API, and managing memory across sessions
The difference between a prompt, a workflow and an agent
Multi-step chains where one output feeds the next
Tool use: how an agent reaches mail, calendar and spreadsheets
Where a human checkpoint is mandatory, and how to avoid harm
Writing a description from scratch and improving an existing one
Shortening it for a different channel
Turning it into a presentation
Generating the questions and the assessment task that follow from it
Building Boolean and X-ray queries from a description
Finding synonyms and the most effective keywords
Writing a first message that gets a reply
Personalising at scale, and the follow-up sequence that lifts response rates
Analysing how attractive your offer actually is
Building an audience portrait, with its pains and wants
Matching the value proposition to that audience
Generating posts, promotion text and illustrations
Welcome communications from both the manager and the organisation
Pre-start questionnaires and the onboarding checklist
The contents of an induction handbook
The 30-day survey, the one-to-one question list and a mentor guide
Building a development plan from a person's profile
Building a competency profile and splitting it into three levels
Designing a path from one role to another
Structuring a training programme and generating the slides
Selecting objectives for any role
Building an assessment matrix, questions and tests
Generating 360 and review questionnaires
Writing feedback after an assessment, and an improvement plan
Building the argument for an idea you need approved
Generating a set of solutions for a business unit's problem
Preparing questions for a difficult situation
Designing a focus group, a facilitation session or an offsite
Designing a bonus scheme for a role and choosing its targets
Selecting factors for job grading
Writing a pay review policy
Designing recognition programmes and referral schemes
Building a culture survey
Formulating mission, vision and values from a description
Turning values into observable behaviours, rituals and artefacts
Preparing an all-hands agenda and the communications around it
Building a strategy template and an audit template
Drafting any policy
Describing any process as step-by-step actions
Building a questionnaire, analysing the responses and generating the report
Generating a metric list and ideas for analysing a process
Analysing turnover and evaluating training effectiveness
Assessing performance, satisfaction and engagement
Calculating the business value of a role
Generating team activity and event ideas
Building the schedule and the invitation
Producing the evaluation questionnaire afterwards
Preparing the exercises themselves
What Claude is and how it differs from the other assistants
The interface and the settings worth changing
Data security and confidentiality when the input is sensitive
Effective prompting here, with two practical exercises
Formulating a request for CV analysis that returns something useful
Preparing questions for a specific role
Building a structured interview
Personalised messages, and rejecting someone without damage
Personalised onboarding plans
Developing training material and programmes
Building a welcome guide
Individual development plans, and adapting content by level
This course contains the use of artificial intelligence.
Most people who say AI is overrated tested it once, typed one sentence, got a mediocre answer, and stopped. That answer was a fair reflection of the sentence.
The gap between using it and using it well
I spent my first year with these tools writing one-line questions and getting one-line quality back. What changed everything was mundane: giving the model a role, giving it context, showing it two examples of what good output looks like, and telling it to check itself before answering. None of that is clever. It is just the difference between a tool that saves you ten minutes a week and one that removes a whole category of work from your calendar.
What this course actually covers
Six blocks, from the basics to agents running unattended. First, the foundation: what these models do, where they stop, and the everyday and work uses that pay back immediately. Second, the craft — the seven blocks of a working prompt, chain-of-thought and tree-of-thoughts for judgement tasks, and the shift from prompting to context engineering with projects, documents and memory. Third, twelve lessons of applied use cases across a whole function: drafting, search, positioning, onboarding material, assessment, reward, culture, documents, analytics and events. Fourth, a second assistant compared on the same work, including documents, spreadsheets, slides and connectors. Fifth, a third assistant inside a document ecosystem, plus writing an internal AI policy and measuring adoption. Sixth, six lessons where each one builds a working agent, ending with several agents orchestrated together and a business case in hours and money.
A note on the examples
Twenty-nine of the thirty-nine lessons use people-operations scenarios: job descriptions, screening, onboarding, assessment, reporting. That is deliberate — it is the function with the densest set of repeated text tasks, which makes it the clearest place to show what these models handle and what they do not. Prompt structure, context engineering, comparing assistants and agent architecture transfer to any function unchanged. The tool names will age faster than the methods; treat them as examples of a category.
Who is teaching this
I am Mike. I built the people system at Preply as it became a unicorn, and I have worked at Wargaming, iDeals and Alfa-Bank. More than 1.6 million students have enrolled in my courses across 185 countries, and over 150,000 specialists have gone through my programmes. I hold PHRi and SHRM-CP certifications and represent HRCI in more than ten countries.
What is included
Lifetime access to all 39 lessons
Active instructor support in the Q&A section
A Udemy Certificate of Completion
Working material: the seven-block prompt structure, four prompt frameworks, context and project setup, a cross-platform comparison, agent architecture
Six agents built step by step, ending with a 90-day roadmap and a business case
Where to start
Take the last thing you asked an AI assistant and read your own question back. Was there a role, context, an example, a format and a constraint in it? If not, you have already found what this course changes. Enrol now and start today.