This week, we begin with AI-Native Developer Workflow.
The goal is to turn a vague product idea into specified, implemented, and independently verified software with coding agents, while keeping requirements, key decisions, and final review in human hands.
You will work on:
- Turning an idea into a product specification and focused backlog.
- Giving agents durable context with
AGENTS.md and project documents.- Separating product manager, software engineer, and QA roles.
- Using loop engineering around checkable completion conditions.
- Coordinating specialized agents across a backlog with graph engineering.
Materials: https://github.com/DataTalksClub/ai-dev-tools-zoomcamp/tree/main/01-ai-native-workflow
Recording: https://www.youtube.com/watch?v=VUJxJGpaDEs
Companion article: https://aishippingblog.com/p/ai-native-development-specifications
Homework 1 asks you to apply this workflow to a spec-driven Django application based on a deliberately vague shared-household-chores idea.
Instructions: https://github.com/DataTalksClub/ai-dev-tools-zoomcamp/blob/main/cohorts/2026/01-overview/homework.md
Submit here: https://courses.datatalks.club/ai-dev-tools-2026/homework/hw1
Deadline: September 8, 2026 at 01:00 CET (September 7 at 23:00 UTC).
The homework is useful practice, but it is not required for the certificate.
Questions belong in the course Slack channel: https://app.slack.com/client/T01ATQK62F8/C09HWT76L95
