If you know SQL, build this agent next

If you know SQL, build this next

(Your next move from SQL reports to AI applications)

Start with a business question, then follow this process

(Read on to discover how👇)

Hi John,

If I were planning my next career move as a data scientist, I’d aim for AI Engineering.

Look at the hiring data.

By the end of 2025, Indeed found:

• Overall tech job postings were down 34% from February 2020.

• Tech postings mentioning AI were up about 45% over that same period.

That’s a reason to think carefully about the skills you’re building next.

Your experience with data, models, and business problems gives you a foundation. The opportunity is to use it to build AI systems that teams can actually put to work.

That’s the move I’ve been calling Track 1 to Track 2.

• Track 1 (Dying): Write the SQL. Build the dashboard. Train the model. Hand over the analysis.

• Track 2 (Growing): Connect the data, models, and AI into a working system that helps a team make decisions.

The danger is getting better at Track 1 while never building anything for Track 2.

Your next SQL project can be your first move across.

So how do you make that transition?

Take one business question you already understand and work through these five steps:

Ask → Generate → Inspect → Verify → Build

Here’s how I’d do it.

1. Ask: Start with a business question

Keep it familiar: “How much did we sell each month?”

Use your Python and SQL skills to understand the data and produce an answer you can check.

Save your usual query and its results. You’ll use them to check the agent’s work.

That gives the project a clear purpose—and gives you a way to tell whether it works.

2. Generate: Give the AI agent a job

Let the user ask the question in plain English.

The agent’s job is to generate SQL and retrieve the relevant data.

Here’s an example I built

My SQL Database Agent App uses the Northwind sample database. Ask a question, and it returns the generated SQL alongside a table of results.

👉 Watch the 12-minute SQL agent walkthrough.

Get the app code here.

Now your SQL knowledge is helping you build an AI application.

See how this connects to a full AI workflow

I’m showing this live on November 11th at 2:00 PM Eastern:

This is exactly what I’ll demonstrate in my next free AI + Data Science workshop.

I’ll take a real data science problem and show you how to move from:

Raw data → ML analysis → AI agent → Recommended actions → Business-ready workflow

Step by step.

On screen.

Not just prompts.

Not just a chatbot.

A working AI + data science system.

👉 See the AI/DS workflow built live — reserve your seat

Now, here’s where your data experience really matters.

3. Inspect: Look at what the agent actually did

Read the generated SQL.

Look at the returned rows.

Which tables did it use? How did it join them? What did it filter out?

Keep the query and results visible in your app so you can investigate an answer.

A confident explanation is a lot less impressive when the underlying query is wrong.

Your SQL background gives you a way to spot that.

4. Verify: Check the business answer

Compare the agent’s results with the report you saved in Step 1.

Check the details that can change the answer:

• Dates: Did it use the right month?

• Joins: Did it count an order twice?

• Definitions: Does revenue include returns?

A query can run successfully and still give your team the wrong answer.

Correct the mistake. Clarify the instructions. Run the check again.

Start with one business question you can verify. Get that working before adding ten more.

This is useful AI engineering practice: getting a system to produce results you can explain and check.

5. Build: Make the agent usable

Give someone a simple way to ask a question and inspect the answer.

The SQL starter app already includes a Streamlit interface you can study and adapt.

As you develop it for other users, work through access permissions, error handling, and deployment.

For your portfolio, show the business question, the working app, and a check that caught a mistake.

Now you have something specific to open in an interview or demonstrate to your boss.

You’ve taken a familiar SQL task and built an AI application around it.

The Python AI Stack You’ll Use

These tools cover the pieces you need to build AI systems that actually do work, not just chat.

• LangChain — Connects LLMs to your data, tools, APIs, and RAG pipelines.

• LangGraph — Turns simple prompts into reliable multi-step agents with memory, retries, and logic.

• OpenAI Models — Give your applications the reasoning and language intelligence behind the workflow.

• Scikit-Learn — Adds traditional machine learning when you need predictions, classification, or scoring.

• Polars — Handles the data preparation and transformation that feeds your AI and ML systems.

• XGBoost — Gives you powerful predictive models for problems like churn, lead scoring, and risk.

The key is learning how to combine them. That’s where you go from experimenting with AI to building production workflows that solve real business problems.

Why the AI Bootcamp Speeds This Up

You could spend 12–18 months piecing an AI roadmap together yourself.

Or you could shortcut it in 8 weeks:

• Live coding with me in Python.

• Build 3–4 portfolio-ready agents.

• Get the AI Engineering Toolkit: 100+ curated libraries (LangChain, LangGraph, OpenAI, Polars, XGBoost, and more).

That’s why, in just the past 6 months, multiple students have gone from “Data Scientist” → “AI Engineer” roles using this exact roadmap.

Final Thoughts: The AI-Driven Future of Data Science

The AI-driven future is here: Routine data science tasks are being automated, but the opportunities for those who adapt are unprecedented: Higher pay, bigger impact, and front-row seats in shaping the future of work.

To thrive, you must evolve:

• From Data Analyst → AI system orchestrator

• From ML model-builder → AI engineer

• From one-off insights → Continuous, agent-driven workflows

Your next step: Start the roadmap today.

👉 Or shortcut the process and join me in the 8-Week AI Bootcamp.

Enjoy!

Your Friendly Neighborhood Generative AI Data Scientist,

Matt Dancho,

P.S. - On November 11th, I have a free AI for Data Scientists workshop that will take your AI skills to the next level. I'll share how you can build an AI Customer Segmentation Agent that automatically performs customer segmentation and provides marketing recommendations for companies (Live Python Code). You can register for free here.

P.P.S. - Cohort 12 of my 8-Week Generative AI for Data Scientists Bootcamp is filling up. As of this morning, 14 of 50 spots are already gone. We start on Monday, February 1st. You can learn about it here.