Your Airflow DAGs run on schedule, dbt transforms the data, and your Snowflake warehouse is humming. On paper, the stack looks modern. In practice, a schema change breaks three downstream models before anyone notices, and fixing it still takes a ticket, a Slack thread, and two engineers' Tuesday afternoons.
Most data teams are further along the automation curve than they were five years ago — and not as far along as they think. Before you dive into ADE 101, take five minutes to map where your team actually stands. Your results tell you which modules to prioritize and where to slow down.
Where Do You Stand? — ADE Maturity Assessment
A five-minute diagnostic that maps your team's readiness across technology, process, culture, and skills — and tells you where to focus in this course.
What your results mean
The assessment maps your team to one of five automation maturity levels — a condensed rollup of the spectrum from manual work through autonomous agents. The full six-level breakdown (Manual → Scripted → Automated → AI-Assisted → Agentic → Autonomous) is in What Is Agentic Data Engineering? (Module 2). This assessment combines the Manual and Scripted bands into a single starting level.
Most teams aren't where they think they are. The gap between "we use AI tools" and "we run agents" is wider than a tool upgrade — it's a systems design problem.
| Level | Label | What it means | Where to focus |
|---|---|---|---|
| 1 | Manual | Most data pipeline work is manual and reactive. Automation is limited to basic scheduling. | Module 2–Module 4 give you the foundation. Module 6 shows the lifecycle view. |
| 2 | Automated | You have scheduled pipelines and basic orchestration. A common starting point for many teams. | You're the primary audience for this course. Start with Module 3 on why the moment is now. |
| 3 | AI-Assisted | Your team uses AI tools daily for code generation and debugging — but not yet running agents. | Module 4 through Module 6 cover the gap from AI-assisted to agentic. Module 7 is your hands-on lab. |
| 4 | Agentic | Agents are deployed in your stack. You align governance (the policies and controls around how agents operate) with how much autonomy agents get by risk tier (the potential impact of an agent mistake in that domain). | Module 6 covers the full lifecycle view. ADE 201 covers system design at depth. |
| 5 | Autonomous | Agents manage significant portions of your data pipeline lifecycle (build, run, and change management) with appropriate oversight. | This course is a foundation refresh. ADE 201 and ADE 301 are your next steps. |
Your answers are not sent to a server. After you submit, your responses are encoded in the page URL — if you share that link, your selections are included. A course completion flag is stored locally in your browser. Screenshot or copy your results to keep them.
⏱ 10 minutes
You have a score — now test whether you can reason about what separates one maturity level from the next. Open any LLM — Claude, ChatGPT, or Gemini work well — and paste this:
You're a data engineering advisor reviewing a team's current stack. Here's the situation:
The team at Expeditions (an outdoor gear company) runs a daily orders pipeline called
orders_daily. It's built on Airflow and dbt, runs on a schedule, and loads results into
Snowflake. When a new data source is added upstream, an engineer manually updates the dbt
models, reruns the pipeline, and checks the dashboard for anomalies.
The team uses GitHub Copilot for code generation and ChatGPT occasionally to debug SQL.
They don't have agents running in production.
Based on a five-level ADE maturity model (1 = Manual, 2 = Automated, 3 = AI-Assisted,
4 = Agentic, 5 = Autonomous), what level is this team at? What specific capabilities would
move them to the next level? Be concrete about what would have to change in the
orders_daily pipeline specifically.
What to notice: Does the LLM place the team at Level 3 (AI-Assisted)? The critical distinction is that AI tools are being used by engineers to assist tasks — not running autonomously in response to events. The response to a schema change still requires a human to trigger it. Watch for whether the LLM names that gap precisely, or conflates "using AI tools" with "running agents."
Your maturity level is a starting point, not a ceiling. Most teams move from Level 2 to Level 3 faster than they expect — the harder jump is from AI-Assisted to Agentic, where tools stop assisting engineers and start operating autonomously in response to events. The rest of ADE 101 is built to prepare you for that jump.
Before agents can help, you need a clear mental model of what they actually are — how they differ from the AI tools your team is already using, and why the architecture underneath them matters.
Next: What Is Agentic Data Engineering? →
Additional Reading
- Data Engineering Trends in 2025: Your Roadmap to Smarter Data Teams (Ascend, 2025) — Directly contextualizes where your assessment score sits relative to where the broader industry stands on automation and AI adoption.
- Building Effective Agents (Anthropic, 2024) — Anthropic's framework for how agent autonomy scales with trust and task complexity; the conceptual foundation for the maturity levels in this assessment.
- State of Analytics Engineering (dbt Labs, 2025) — Broad benchmark of where analytics teams are on automation and AI adoption today; useful context for Level 2–3 teams mapping their position relative to peers.
- Stack Overflow Developer Survey 2024 — Annual data on AI tool usage and adoption among engineers; shows how quickly "AI-Assisted" has become the baseline and what Level 3 teams now look like across the industry.