🟒 A Practical Map for Using AI in Analytics


Hello Reader,

This week's newsletter is different. No SQL code. No database queries.

I lead a team of 30 data professionals, and we use tools like Claude and Copilot to get oriented, draft, review, and explain routine analysis. The useful part is moving faster without handing over the judgment your team relies on.

We call this "human in the loop".

Agentic Analytics means using AI as part of your analytical process: give it context, check its work, and keep responsibility for the decision.

I remember building an early pipeline report from SQL that a teammate had to run for me. The Excel report helped, but I could not explain or improve the logic underneath it. AI gives you a better starting point today. You still need enough understanding to direct the work and stand behind it.

That's what we've been covering each week in this newsletter for the past several months:

But over the next several weeks, I will share practical agentic analytics methods with Summit Adventures examples you can use to level up your own skill. Your own AI Leverage framework -- there are five levels.

The 5 Levels of AI Leverage

Each level builds your skill in a practical way.

  • Level 1: Learn with AI. Use it to explain an inherited query, decode errors, or clarify something you don't understand.
  • Level 2: Draft with AI. Let it write a first pass at a query or analysis, then check it before anything ships.
  • Level 3: Analyze with AI. Use it to explore angles, pressure-test an interpretation, and draft an executive summary.
  • Level 4: Automate with AI. Turn a recurring analysis into a scheduled workflow
  • Level 5: Monitor with AI. Set an alert around a decision-relevant threshold, such as dormant high-value customers needing a retention action.

I'll admit: this is all BRAND new stuff. But I'm seeing it crystalize basically every week in real time. And that's what I want to share here.

Your Expertise Still Matters (More Than Ever!)

AI can EASILY draft a query.

It will probably run but will likely miss something along the way.

  • A join that accidentally drops customers who never booked.
  • A filter can exclude the rows that matter most.

This is because AI doesn't have all of the business context required. That's where you come in.

By keeping you in the loop, you catch those problems before the results are published.

The habits you have built by reading and applying in the Analytics in Action newsletter still work:

  • explore the data
  • look for what is missing
  • and interpret the numbers for a decision

AI can speed up the work but you still have to know what's going on under the hood.

Where This Series Is Going

Over the next several weeks, each newsletter will teach one method you can use on its own, with examples from Summit Adventures, the fake adventure tourism company I created to help people learn business analytics.

Highlights include:

  • The Draft-and-Check Method: the four-step way to let AI write SQL without losing control of quality
  • The briefing document: a reusable way to give AI useful context
  • AI-assisted exploration, executive summaries, and data quality audits
  • reusable workflows, automation design, and monitors

This Week's Action Item

Pick one report or analysis you repeat. Write down the decision it supports and finish this sentence: "If this ran at Level 4 or 5, it would ___ without me."

Send me your thoughts on this -- I wanted to start a new series since we've been covering SQL in depth for the past few months. What Level would you currently say you are at? Are you using AI in your daily work at this point? Hit reply and let me know.

Until next time,

Brian

Brian Graves, creator of Analytics in Action

Say πŸ‘‹ on X/Twitter, LinkedIn, or book a call with me. You can always reply to these emails. I check them all.

​

Starting With Data

Learn to build analytics projects with SQL, Tableau, Excel, and Python. For data analysts looking to level up their career and complete beginners looking to get started. No fluff. No theory. Just step-by-step tutorials anyone can follow.

Read more from Starting With Data

Hello Reader, Real quick: if you haven't completed this subscriber survey -- now's your chance. Most people are looking to learn more about Agentic Analytics and AI topics. I want to build something that would be useful to you, based on the work I've been doing for the past year or so. Interested? Take 5 minutes to let me know what's on your mind. Take the 5-Minute Survey Now πŸ‘‰ Imagine it's your first week at a new job. Someone from IT sends over database credentials and a two-line email:...

Hello Reader, Happy 4th of July to my fellow Americans -- glad you're with me today. πŸŽ‡ I'm planning out newsletters for the rest of the year and want to start incorporating more Python, Tableau, and Agentic Analytics. Can you take this short survey for me? Thanks! If you've been working as an analyst, you'll be familiar with a request like this: "This weekly report is exactly what we needed. Can we get it every Monday morning, with a chart, emailed to the whole leadership team automatically?"...

Hello Reader, A while back, I shared some tips about formatting SQL output for spreadsheets. Today we're going the other direction. Doing the kind of analysis that makes spreadsheet formulas feel clunky. Here's the kind of request that analysts get every single day: "Show me monthly revenue and how the trend is moving." If you pull monthly revenue into a spreadsheet, you'd probably create a running total column, then a 3-month moving average column, then a month-over-month change column....