AI-Native Product Design: From Job to Safe Launch

What's inside

1. Level 1: AI-Added Versus AI-Native ProductsFree
   Learn the structural differences between products that bolt AI onto an existing workflow and products redesigned around what AI makes newly possible.
2. Level 2: The Running Case: A Regulated Service WorkflowFree
   Meet the course's running case, an AI-native product for a regulated service industry, and describe its users, the jobs they need done and the constraints they work under.
3. Level 3: Picking the Job: Where Models Are Strong or Weak
   Break the case's jobs into tasks and score each one on model strengths, failure risk and outcome value to choose the first job to automate.
4. Level 4: Mapping the Current Workflow
   Map the chosen job's workflow from start to finish, covering steps, hand-offs, inputs, decisions and points of failure, before redesigning it.
5. Level 5: Dividing Work: AI, Person and Check
   Redesign the workflow by assigning each step to the AI, to a person or to a verification check, and record the reason for every assignment.
6. Level 6: Autonomy Levels: From Copilot to Agent
   Define a scale of autonomy levels, from suggestion-only copilot to independent agent, and choose the right level for each task in the case.
7. Level 7: Interfaces Beyond Chat
   Choose interface patterns that fit each task, including generative UI, inline suggestions, background agents and review queues, instead of defaulting to chat.
8. Level 8: Designing for Errors: Confidence and Citations
   Design how the product shows uncertainty and evidence so that users can tell which outputs to rely on and which to check.
9. Level 9: Recovery: Undo, Correction and Escalation
   Design the recovery paths for when the AI gets something wrong: undo, inline correction, escalation to an expert and audit trails.
10. Level 10: Trust and Transparency
    Help users build calibrated trust by being clear about what the AI did, what it can and cannot do, and how decisions were reached.
11. Level 11: Evaluation as a Product Feature
    Build evaluation into the product with eval sets, golden tasks and regression tests that decide whether a change can ship.
12. Level 12: Data Flywheels and Feedback Loops
    Design feedback capture and data loops that turn everyday use into better evals, prompts and models without adding friction for users.
13. Level 13: Memory and Personalisation with Privacy
    Decide what the product remembers about users and clients, for how long and under what controls, so that personalisation does not compromise privacy.
14. Level 14: Cost, Latency and Model Choice
    Treat cost per task, response time and model selection as design constraints that shape the workflow and the interface.
15. Level 15: Safety, Security and Prompt Injection
    Outline the main safety and security threats to an AI-native product, including prompt injection, data leaks and misuse of the AI's tools, and the design controls that limit them.
16. Level 16: Regulation in Outline: EU, UK and Sector Rules
    Get an overview of how the EU AI Act's risk-based approach, the UK's regulator-led approach and sector-specific rules affect design decisions in a regulated service.
17. Level 17: Measuring Value Delivered
    Define and measure the outcomes that show the AI-native product delivers real value, beyond usage and engagement figures.
18. Level 18: Prototyping Fast with AI Tools
    Use AI prototyping tools to build and test workflow, interface and prompt prototypes quickly, with realistic but safe test data.
19. Level 19: Staged Launch and the Design Review Checklist
    Plan a staged, safe launch with defined gates, monitoring and rollback, and run a full design review checklist before each stage.
20. Level 20: Final Challenge: End-to-End AI-Native Design Exam
    Apply the whole course to a new regulated-service scenario, taking it from job selection to a safe launch plan, and defend each design decision.

Access

The first 2 levels are free with a free account. Every level, the podcast edition and the AI tutor come with All Access at £4.99/month or any Creator plan — see pricing.

Start free →