Become an AI Product Manager. Combine proven best practices with AI know-how to succeed.

What has really changed in a PM's responsibilities, which toolkit you work with today, and how products are built once the iteration loop gets shorter.

• How have the PM role and work changed in the AI era?

  • Definition and differences between similar roles (project manager, product owner, etc.)
  • Classic PM responsibilities vs. responsibilities today: what disappeared, what got automated, what gained importance
  • New competencies: prompt literacy, assessing the quality of AI output, working with uncertainty
  • A shorter iteration cycle: hours/days instead of sprints
  • The PM as an orchestrator of AI and human work
• The AI PM toolkit

  • Research and synthesis tools (ChatGPT, Claude, Perplexity)
  • Prototyping tools (Lovable, Claude Code, v0, Bolt)
  • Tools for data analysis and automating PM work
  • How to choose a tool for your stack
• How products are built now

  • A shorter loop: hypothesis, prototype, test
  • The PM as a “one-person prototype factory”
  • The changing role of engineers: from building to review and architecture
• Examples of AI tools

  • Demo: the same PM process done the classic way vs. with AI
  • Case studies of companies that changed their process thanks to AI
  • Exercise: testing 2-3 tools on your own example

Discovery when users haven't seen the product yet, and “a lot of data” doesn't mean “data you can use.”

• Discovery in the AI era

  • A shorter discovery cycle thanks to AI
  • New signal sources: conversations with agents, telemetry from prototypes
  • The PM as a curator of insights
• Researching the need for an unknown product

  • Limitations of surveys and interviews for AI-native products
  • Wizard-of-Oz testing, concierge MVP
  • Observing behavior instead of declarations
• “A lot of data” is often useless data

  • Data quality vs. quantity, common pitfalls
  • Legal constraints and privacy
  • AI for organizing unstructured data

Sound product decisions and arguments that hold up in front of the board.

• Making sound decisions

  • Framework: data + intuition + business context
  • Reversible vs. irreversible decisions
  • AI as a sparring partner for testing decisions
• Defending a product/feature

  • Mapping stakeholders and their motivations
  • Narrative: problem, evidence, solution, impact
  • Managing expectations and objections
  • Handling escalations and tough questions
  • Jobs to Be Done
• Talking to the CFO/CEO

  • Business language: ROI, opportunity cost, North Star metric
  • A business case with the help of AI
  • Presenting data on a single slide
  • Exercise: a 3-minute pitch justifying a feature

The roadmap, priorities, and the road from idea through POC and MVP to product-market fit - tied to the vision and business goals.

• Building a roadmap

  • Roadmap types: outcome-based, feature-based, now-next-later
  • Connecting the roadmap to the vision and strategy
  • Communicating the roadmap to different audiences
  • More frequent roadmap reviews thanks to faster feedback loops
• Product management

  • Vision and strategy as the reference point for decisions
  • SWOT as a tool for periodic audits
  • Managing the backlog for value
  • The product lifecycle and decisions at each stage
• Prioritizing initiatives

  • Frameworks: RICE, Value vs. Effort, Kano
  • Prioritization based on metrics/KPIs
  • Decisions with incomplete data
• The road to production

  • Stages: idea, market analysis, POC, MVP, PMF
  • Signals of real vs. false PMF
  • The role of segmentation and cohorts in confirming PMF

You set up the terminal from scratch and teach an AI agent to work with your product's context.

• Building product context

  • What “context” means for an agent (PRD, data, communication style)
  • File and knowledge structure for the agent
  • Context as a living, continuously updated document
• Launching and configuring the terminal

  • Terminal basics for PMs without a tech background
  • Installing and configuring a tool (e.g., Claude Code)
  • Basic commands and navigating a project
• How to get started with a product?

  • Choosing a first, small project
  • Defining the MVP scope for your own tool
  • Examples: a reporting assistant, a PRD generator, a documentation bot
• Working with an agent

  • Formulating tasks and instructions
  • Iterating and improving results
  • When to take control and when to trust the agent
• Security

  • Sensitive data: what not to paste into an agent
  • Permissions and sandboxing
  • Verifying results before deployment
  • Compliance with company policies

From PRD to a clickable prototype - with your own hands, without involving the IT department.

• Writing a PRD

  • The structure of a good PRD
  • AI speeds up the draft, user stories, and acceptance criteria
  • The PRD as input for the agent building the prototype
  • Communicating design requirements to the team
• A prototype without IT (Lovable, Claude Code)

  • Overview of no-code/AI-code tools
  • From idea to clickable prototype
  • Limitations: what you can build without engineers
• Working with LLMs while prototyping

  • Iterative prompting: from general to specific
  • Debugging and fixing results
  • Keeping the product consistent across iterations

You define what a good answer means, build evals, and choose the right model for the task - with costs on the table - before the product reaches users.

• What does a good answer mean?

  • Quality criteria: accuracy, tone, safety, factuality
  • Evaluation rubrics and who does the evaluating
  • “Works technically” vs. “works for the user”
• Building evals

  • Golden dataset and metrics (accuracy, hallucination rate, satisfaction)
  • Continuous evaluation when the prompt/model changes
  • Evals as the basis for release decisions
• Choosing models and architecture

  • Which model for which task - quality vs. speed vs. cost
  • Self-host vs. API (Claude and others) - when each makes sense
  • Costs: tokens, infrastructure, hidden scaling costs as usage grows
• Pre-release testing (condensed)

  • Automating regression tests with the help of AI
  • What makes testing AI/LLM features different (non-determinism)

After launch: faster analysis, a dashboard, and A/B tests that tell you whether the product really works.

• How AI speeds up analysis

  • Natural-language queries to data
  • Automatic summaries and anomaly detection
  • When to focus on manual analysis
• Building a dashboard

  • Choosing metrics/KPIs
  • Executive vs. operational dashboard
  • AI for building and adapting a dashboard faster
• A/B tests and other concepts that matter for a release

  • A/B testing methodology basics
  • What makes testing AI features different (prompts, models, qualitative metrics)
  • Analyzing results and making decisions
  • Pitfalls: false significance, tests that are too short
• Recording user interactions

  • Session replay and pattern detection tools
  • Analyzing user-agent conversations as a source of insights
  • User privacy and consent