Why AI Skills + Data Foundations Are the New Power Combo at General Assembly

Artificial intelligence is reshaping the way data professionals work – and bootcamps like General Assembly have to move quickly to keep pace. Longtime General Assembly instructor Candace Pereira-Roberts, who is also a Director of Data Strategy and founder of ThoughTrails when she’s not in the classroom, has been on the frontlines of that shift. She helps students build solid foundations in SQL, Excel, and Tableau while also showing them how to layer in generative AI tools to accelerate workflows. Candace shares how GA is weaving AI into its data curriculum, why foundational skills still matter more than ever, and her advice for anyone considering a career in analytics today.

Candace, you’ve been a Director of Data Strategy and a founder – what brought you to teach data analytics at General Assembly?

I actually found General Assembly before I ever managed a team – I was looking for a data science course for myself. What drew me to teaching here was the importance of sharing knowledge. GA gives people solid, hands-on foundations to launch a career in analytics or data science, and I wanted to be part of that. Analysts often struggle to get the right skills, and GA provides a path. Teaching allows me to watch people grow, which I love. Today, I primarily teach data analytics and generative AI – where AI meets analytics and data science. That’s what keeps me coming back. I’ve been with GA for about eight years now, and I really love to watch people grow.

What's your favorite class to teach right now?

I love teaching the generative AI course because it’s so new and exciting. But when it comes to core analytics, my favorite is definitely SQL. It’s my wheelhouse – I speak multiple dialects of SQL – and it pairs perfectly with AI. So I’d say AI and SQL together are my favorite combo to teach.

What are you doing when you're not teaching?

I came up through the ranks before moving into leadership. I’ve worked as a BI analyst, data warehouse specialist, technical writer, and data engineer. That technical writing role is actually what sparked my interest in data. Today, I manage a team of analysts and data scientists as a Director of Data Strategy. 

You've seriously done it all! So how do you bring all that real-world experience into a workshop or classroom?

I’m known for bringing clarity – professionally and personally. I like to break concepts into small, digestible pieces. My mantra for students is: “The only way to eat an elephant is one bite at a time.” My teaching style mirrors GA’s model: show, guide, practice, and recap. I also use analogies to connect new concepts to familiar ones. For example, I’ll compare SQL’s “GROUP BY” to using rows in a pivot table in Excel – it bridges the gap for students.

The students in my classroom have very diverse backgrounds. Some are great at Excel, others have worked with SQL or visualization tools. Others are totally new. But if you just run your own race, you'll get there. Everybody gets out of it the effort they put in. I think I learn from my students as much as they learn from me.

For someone considering a data analytics career, why is it important to learn AI skills right now?

It’s not just in analytics – AI is impacting everything. But AI isn’t replacing us; it’s a new technical revolution, much like when Excel came on the scene in the ’90s. At first, people resisted, but eventually Excel became essential. The same is happening with AI. Those who learn it and get comfortable with it will be more successful long term. It’s everywhere already, so building confidence with these tools is key. And remember: you can’t break it.

For example, we’ve seen the headlines that SQL will be replaced by AI. But you’ve said GenAI + SQL is a “power combo” – what does that look like in practice?

Generative AI is like an intern – helpful, but it makes mistakes. I often ask it to write SQL, and sometimes it aggregates data incorrectly. Because I eat, live, and breathe SQL, I can catch those errors. The best use of AI is optimization: it can suggest improvements or give you a starting point, but you need foundational knowledge to validate the output. Humans will always be required in the driver’s seat. That’s never going away, in my opinion. Gen AI is just helping you optimize your workflow, not replace it.

Where do you see AI already working well in analytics?

Recently, I used AI to generate Python code for visualizations in an impact analysis project. It saved me time and helped me learn new techniques. I still had to check the results and ensure the charts told the right story, but AI expedited the workflow and gave me “aha” moments along the way.

You’ve been at General Assembly for almost years. How are you seeing AI incorporated into the data curriculum these days? How much is GA changing with the times?

GA is definitely keeping up with the times. There’s now a module on generative AI for analytics in the data analytics course. It’s a flexible session – instructors can choose to include it – but I think it’s essential. I usually teach it midway through the course, so students build their foundations first and then layer AI on top.

Beyond the core courses, GA is also rolling out short, three-hour workshops like AI for Analysts or AI for Marketers. These context-specific sessions help professionals see exactly how AI applies in their field, whether analytics, marketing, or finance. Prompt engineering looks different for each role, so tailoring the curriculum makes it more relevant.

On top of that, GA is updating its teaching around traditional tools as they add AI features. For example, Tableau now has a “Talk to Data” feature, and many platforms – from databases with Copilot to visualization tools with Gemini – are embedding AI assistants. Teaching students how to use these built-in features is essential, not just for efficiency but also for security. If the AI is embedded in a company-approved platform, the organization can vet it and ensure sensitive data isn’t exposed publicly.

So that’s more context-specific – AI for marketers, AI for legal, AI for medical. It sounds important that it gets tailored to the role people are already in.

Exactly. Tailoring it ensures students learn what’s most relevant to their work. Prompt engineering – just knowing how to interact with AI tools like ChatGPT or Gemini – looks different for an analyst than it does for someone in finance. GA designs these sessions so the examples and use cases fit the audience.

Where should students or new analysts be cautious with AI?

First, only use the tools your organization approves – don’t upload sensitive company data into public tools like ChatGPT. Second, ethics matter. AI should not replace a doctor’s diagnosis; for example, it should be a research aid. Always validate results with your own expertise. Use public datasets from places like Kaggle, the World Health Organization, or other government sites for practice. And if you’re brand new, you can even ask ChatGPT to generate sample datasets for you.

That's really good advice. Could you share one or two examples of real-world problems that students might solve with AI in the courses that you're teaching? 

In the data analytics course, students complete a hypothesis-based project. Traditionally, that means creating a hypothesis, developing supporting questions, and then using data to prove or disprove it. The datasets we use at GA are fictional but realistic, so students can practice without worrying about sensitive information.

This is where AI becomes a useful assistant. For example, students might use it to:

1. Brainstorm project ideas: Drop a dataset into an AI tool and ask it to suggest hypotheses and supporting questions. That way, students spend less time getting stuck on “What problem am I trying to solve?” and more time analyzing the data.

2. Optimize SQL workflows: AI can write or refine SQL queries, suggest code snippets you can adapt to your own database, or streamline repetitive steps in the analysis.

3. Translate data into insights: AI can help turn technical results into clear, actionable business recommendations for stakeholders.

4. Plan and organize projects: In broader contexts, AI can take multiple stakeholder requests, organize them into a roadmap, and set realistic timelines and deliverables.
   
   

All of these examples reflect real-world scenarios. In the workplace, a business leader might come to you and say, “Why is profitability down year over year?” and it’s your job to dig into the data and find the root cause. AI can’t replace the analyst, but it can speed up workflows, remove roadblocks, and sometimes even introduce automation — allowing you to focus on the higher-level analysis.

Which tools are the most important to prioritize in 2025/2026? What’s on its way out? 

The big players – ChatGPT, Copilot, and Gemini – will be around because they’re backed by major companies and keep improving. There are new entrants like DeepSeek, and beyond that, there are niche tools like Grammarly or BeautifulSoup for presentations. But remember: many tools are interfaces using the same large language models under the hood. The best advice is to try them, see which interface works for you, and stick with what fits your workflow.

And as you are mentoring the next generation of data folks and data analysts, how do you think about balancing teaching AI skills, these kinds of, you know, front-of-the-pack skills, with building that solid foundation in traditional analytics tools?

You can’t build a house on sinking sand. Data Analysts need solid foundations – SQL, visualization tools, maybe some Python – before layering AI on top. Data scientists should build a foundation in Python, statistics, and math.

I also warn students not to “copy-paste blindly.” AI can hallucinate, generating false URLs or made-up facts. I call it a cocky intern: useful, but it needs supervision. Foundational skills let you validate and challenge AI output.

Who are your courses for – do you need a tech background to succeed? Who will get the most out of these courses?

It’s not about technical background – it’s about effort. Some of my best students had zero experience but worked consistently and produced amazing projects. Meanwhile, technically strong students sometimes procrastinate or coast. Success really depends on your work ethic and persistence.

If you aren’t an analytical thinker or unwilling to dedicate time and effort, these courses aren’t for you. This isn’t a “bird course” you can just fly through. If you’re curious but unsure, try a short workshop first. GA offers free two-hour workshops that let you dip your toe in before committing to a full 10-week or immersive bootcamp.

You often say, “start before you feel ready” – how does that mindset help someone learning AI and analytics?

Absolutely. If you wait until you feel ready, you’ll never start – fear will hold you back. You don’t have to dive into the deep end, but at least get in the pool. Every journey starts with a single step. AI was new to me once, too. You just have to try, practice, and improve over time. And yes – eat the elephant one bite at a time.

Find out more and read General Assembly reviews on Course Report. This article was produced by the Course Report team in partnership with General Assembly.