Making AI Work in Data Teams: A Practical Guide
Create an Environment Where AI Adoption Flourishes.
Let me guess: As a team lead, one of your 2025 goals is to help your team harness AI tools to work more efficiently. With the overwhelming pace of new developments and varying comfort levels across your team, it's hard to know where to start. Here's the reality: successful AI adoption grows organically—through experimentation, practical applications, and peer learning. You can't force it, but you can create an environment where it flourishes naturally. Here's how to make it work in your team.
Creating the Right Environment
Encourage experimentation
Adopting AI tools and workflows takes time. Like any skill, you need to familiarize yourself with new tools, and AI is no different. What sets AI apart is that there's no clear playbook for using it effectively. While there are best practices for prompting and guidelines about AI's strengths and limitations, you need to discover what works for you. AI isn't just another programming language or company tool—it's a completely new way of working, and we're all beginners in this revolution. Finding what works best requires experimentation, with both successes and failures along the way. You'll be amazed by AI's capabilities in some moments and puzzled by its limitations in others. Ethan Mollick calls this the "jagged frontier"—an invisible border between what works and what doesn't. This frontier varies not only by domain but also by individual, as everyone has unique needs and contexts.
To foster AI adoption, give your team time and space to experiment. No one fully understands what this technology can do when starting out, and unlike programming languages, there's no structured learning path. Spark curiosity in your team, remove friction (e.g., by providing company licenses for ChatGPT), and keep the conversation going. Once your team starts experimenting, you're halfway there.
Set expectations
While generative AI is rapidly evolving with each new model release, setting realistic expectations has been key to successful adoption in our team. Even as capabilities grow, understanding AI's limitations is just as important as exploring its potential.
To create an AI-friendly environment, we first ensured everyone understood the basics: how large language models work, their common pitfalls, and fundamental prompting techniques. When my team grasped how LLMs work, they better understood both the impressive capabilities and occasional bizarre responses. This knowledge transformed AI from either a magical solution or an unreliable gimmick into what it truly is: a powerful but imperfect collaborative tool.
Understanding these fundamentals makes teams more resilient in their AI journey. Rather than giving up when hitting limitations, people learn to find creative workarounds. They start viewing prompting as a skill to develop rather than a fixed rulebook to follow. Setting these expectations early creates an environment where both success and failure become valuable learning experiences, helping to make AI adoption an exciting journey of discovery.
Practical Strategies That Work
Low entry points
A common mistake when starting with generative AI tools is thinking too big. It's natural—everywhere you hear about how AI is changing work and replacing jobs. In practice, the use cases are often more mundane. Start small. Don't try to run right away; take baby steps. Familiarize yourself with the technology before attempting to revolutionize your workflow.
My team initially tried using AI tools to generate SQL queries, thinking it was the obvious use case. Since ChatGPT excels at writing SQL, we assumed applying this to our data warehouse would be straightforward. It turns out Text-to-SQL remains challenging, and our simple attempts fell short. While I don't regret trying this approach, focusing on smaller, achievable tasks is more productive. Encourage your team to use AI tools to review existing queries, explain query logic, or structure new queries. As mentioned earlier, ensure everyone has reasonable expectations. When talking to colleagues and friends, these use cases are good entry points for beginners:
Assisting with communication: summarizing emails, refine tone of emails.
Instead of googling a question that would inevitably lead to Stack Overflow, ask ChatGPT about the problem.
Ask ChatGPT questions about a PDF.
Build Momentum Through Champions
The AI space evolves at a dizzying pace, with new tools and capabilities emerging weekly. As a manager, part of my role is curating these developments and identifying which ones could genuinely benefit the team. But introducing new tools effectively requires more than just announcing their availability—it needs champions who can demonstrate their real-world value.
A powerful example of this dynamic played out recently when we introduced Cursor, a new AI code editor, to our team. While many developers had grown skeptical of AI coding assistants after mixed experiences with GitHub Copilot, one of our data engineers saw Cursor's potential immediately. He began using it in his daily work, showing others how it significantly improved his workflow during code reviews and pair programming sessions. His authentic enthusiasm and practical demonstrations did more for adoption than any formal introduction could have achieved.
This pattern of champions naturally emerging to drive adoption has become a cornerstone of our AI integration strategy. For instance, one of our data analysts became our unofficial "ChatGPT whisperer" after extensively using it for stakeholder communication. He showed the team how to draft polished responses, streamline reports, and even review SQL queries. These weren't just demos—they were real work problems being solved more efficiently. His success sparked discussions about query efficiency and inspired others to experiment with similar techniques.
The key is recognizing that new technology adoption and champions form a virtuous cycle:
Champions help evaluate and validate new tools in real-world scenarios
Their successes make adoption feel more achievable for others
This creates space to introduce more advanced tools
Which in turn empowers champions to push boundaries further
As a manager, your role is to identify these champions early and support them. When you spot a promising new tool or technology, share it first with team members who have shown enthusiasm for AI experimentation. Their practical experience and peer influence will help determine if and how to roll it out more broadly.
Remember: while countless new AI tools emerge daily, the goal isn't to adopt everything. Be selective and let your champions help guide which tools truly add value to your team's specific needs and workflows. Their ground-level perspective is invaluable in cutting through the hype and identifying practical applications that stick.
Managing The Human Side
As with any change, some individuals will embrace working with AI while others will be more reserved. To drive adoption, it's crucial to ensure everyone is aligned and ready to embark on this journey. When introducing AI tools, these approaches helped me tremendously:
Hands-on Workshops & Knowledge Sharing
One effective way to spark curiosity is through hands-on workshops. These sessions provide a safe space for team members to share their experiments with AI tools. They can showcase their successes and discuss their failures. Both are invaluable lessons. Sometimes a failure can even be corrected (e.g., by improving the prompt), which in turn sparks interest in mastering these technologies.
An important aspect of these workshops is that they shouldn't be lectures. Nobody learns by being told how to do things. Make it hands-on, interactive, and collaborative. These workshops should feel more like kindergarten than university.
Once your team embraces the change, encourage them to share their experiences so others can learn. While finding the right way to use AI is personal, the lessons learned are valuable for everyone.
Success Stories
One way to boost AI tool adoption is through friendly competition. In my team, we held a mascot competition where everyone could contribute suggestions for our team mascot—with one condition: it had to be AI-generated. Though image generation isn't typically relevant for data teams, it encouraged participants to think about AI interaction and playfully improve their prompting skills while learning about GenAI's quirks (e.g., six-finger hands, made-up text in images).
The winner received a book, and we used the mascot as a logo everywhere possible: our stakeholder Teams chat icon, JIRA board, and PowerBI workspace. It even made its way onto T-shirts we gifted to departing colleagues.
Conclusion
The journey to AI adoption in your data team doesn't have to be overwhelming. By creating the right environment, starting small, and empowering champions, you can transform AI from a buzzword into a practical tool that genuinely enhances your team's capabilities. Remember that success isn't measured by how many AI tools you adopt, but by how effectively they solve real problems and improve your team's work.
The key is to maintain balance: embrace the technology's potential while acknowledging its limitations, encourage experimentation while setting realistic expectations, and most importantly, keep the focus on your team's growth and development. As you move forward, celebrate the small wins, learn from the setbacks, and keep fostering an environment where AI adoption can grow organically.
Your role as a leader isn't to force AI adoption, but to nurture an environment where it can flourish naturally. By following these principles, you'll find that your team doesn't just adapt to AI—they thrive with it.