Future Lab: AI Coding

• Compare how differently agents respond to a general request and a specific work assignment
• Continue to develop your learning journal on your own

You will methodically select a manageable B2B use case and refine your idea, including by

• Selecting and Describing Your User Group
• The Definition of Triggers and Outcomes
• An Analysis of Three Comparable Solutions
• Validating Your Idea Through Peer Discussion

You will continue to develop your use case idea—which was refined and validated during the second lab session—on your own.  

These include: 

• Your One-Pager 
• The Mood Board (designs) 
• Wireframes or mockups, if applicable 
• The Architectural Sketch/md. 

A structured interview with a language model helps you identify unstated assumptions. 

The hands-on phase concludes with you 

• Specify your product context
• Export your current prototype
• Check the technical status based on “Document 0”

The complete technical context package (Agents.md, architecture plan.md, etc.) will be developed during the next, third hands-on phase.

You'll learn the process for your project:

• Migrate to the work environment where coding agents work with your repository
• Back up the project status using Git and continue developing it in a structured manner 
• Organize work packages in a PLAN.md file; describe individual features with clear boundaries and acceptance criteria  
• AGENTS.md or CLAUDE.md: Define specific working rules for your coding agent
• Use skills, hooks, and MCP where they simplify a specific task

In addition, you'll learn the basics of controlled loops and tests. 

You create the work packages (context package) for your own use case:

• PLAN.md  
• Individual features with clear boundaries and acceptance criteria
• AGENTS.md or CLAUDE.md: Work Guidelines for Your Agent
• Skills, Hooks, and MCP: How They Simplify a Specific Task

In doing so, you'll apply the fundamentals of controlled loops and tests.

In the firstchallenge, you'll submit your context package (from Hands-on Phase 3) and receive written feedback on your understanding of the problem, scope, assumptions, and testability.  

Only after this feedback is received does the implementation begin using the newly generated repository code. 

Before the next lab session:  

Loop and Graph Engineering: Controlling the Behavior of Coding Agents: 

• Within an implementation window of about three days, you'll build the first working version of your project—your MVP
• In doing so, you'll apply the fundamentals of controlled loops and tests

• Iterating on Your MVP: Specifying Requirements, Delegating Tasks, Reviewing Results, and Learning from Them
• Link acceptance criteria to appropriate tests
• Test-Driven Development as the Foundation for Verifiable Delegation
• Builds, Linting, and Tests: Check Changes and Behavior Before Accepting Them
• Recognize when to stop a loop and first correct requirements, context, or agent rules
• Peer Review: Demonstrate the core workflow, describe the associated requirements and a failed loop, identify pitfalls, and define useful work patterns for the next project phase

• Continue to develop the MVP based on small requirements and systematically review changes
• An Overview of Hosting, Deployment, and the Separation of Development, Test, and Production Environments
• Implementing HTTPS, Authentication, Roles, and Secure Management of Access Credentials
• Assess the importance of backups, logging, CI/CD, and monitoring for your own project
• Describe your application's data processing, develop a data deletion strategy, and test its implementation using synthetic or approved, non-sensitive data
• Cost Control: Set token limits, monitor costs, and review which content is transmitted to model providers
• Record open tasks in the prioritized production readiness plan

Document and present results, and determine next steps

Developing Your Product Evidence Package

• It contains your working prototype
• The repository or a complete export
• Easy-to-Follow Getting Started Guide
• Implemented Requirements
• Test Results
• A Security and Cost Review

Submit this package as Challenge II.

In the secondchallenge, you'll put together your Product Evidence Package and submit it.

It contains:

• Your working prototype
• The repository or a complete export
• Easy-to-Follow Getting Started Guide
• Implemented Requirements
• Test Results
• Security and Cost Review

After receiving written feedback from the instructors and before the final lab session: 

• Targeted revision of the points that provide the greatest benefit
• Preparing for your structured use case pitch in the final lab session:
  • Pitch and a reproducible demo 
  • Explain Important Decisions
  • Explain a corrected loop and known limitations
  • Identify the most sensible next step for your project—for example, a user test, an integration, or technical hardening

Use Case Pitch: 

• Structured Project Presentation 
• Reproducible Demo 
• Explain key decisions, a corrected loop, and known limitations 
• Identify the most sensible next step for your project—for example, a user test, an integration, or technical hardening

Investing Game: 

In the Investing Game, you allocate fictional budgets to the other players' projects. The focus is on demonstrable benefits, solid evidence, and a plausible plan for future development.