You’ve rolled out a modern data stack, built self-service dashboards, and empowered your team to ask their own questions. Job done, right?
Not quite. The data landscape is shifting rapidly beneath our feet, which makes it critical to understand how to build your AI for BI platform so you can scale and navigate technology evolutions. I had the pleasure of sitting down with Jane Smith, ThoughtSpot’s Field Data & AI Chief Officer (EMEA), for a fireside chat at the CDO Retail Exchange in London on The Power of Trust + Context for Agentic Analytics.
We talked about what comes after self-service analytics: the transition to AI-driven, agentic analytics. At Huel, this journey hasn't just changed how we interact with numbers; it has fundamentally redefined the role of my data team and how we drive commercial growth.
If you missed the session, here is my insider blueprint on how a fast-moving consumer brand scales AI analytics without a Silicon Valley budget.
1. Being "Commercial-First" Beats Having a Tech Budget
Huel is a food company, not a tech conglomerate. We are never going to have a Spotify or Netflix-sized engineering budget, and our revenue isn’t built on SaaS-style EBITDA. We have to be commercial-first, which means every piece of tech we deploy must directly drive business growth.
The good news? You don't have to invent the next frontier Large Language Model (LLM) to make massive progress. A scrappy, commercial-first mindset will always lead.
2. Why GenAI Alone Isn't Enough: The Power of Explainability
People often ask me: "Why don't you just give your employees access to ChatGPT or Claude and call it a day?"
We actually tried that. We always want to meet people where they are, so we initially attempted to build models in Snowflake and surface them directly via Claude. What we realized is that standard LLMs act merely as a middleman. They lack transparency.
Our journey split into two distinct eras:
BC (Before ChatGPT): We have a lot of early adopter tech users on our team so as soon as ChatGPT came out they were ready for jump full force into ChatGPT an Claude.
AC (After ChatGPT): We needed to guide our team to see how Spotter wins for us with Explainability and the feedback loop and also to the business compliance aspect to protect our business.
As Huel grows and becomes truly omni-channel, the business questions we face get vastly more complex. To handle this, I've led our team to be "Spotter-first" (ThoughtSpot's AI-powered analytics agent), and we are seeing great success.
With Spotter, our business users don’t just get a static answer; they get complete explainability behind how the data was calculated. More importantly, our users can apply their own unique subject matter expertise to audit and edit data visualizations directly within the platform. If something looks off, they can tweak it in real-time. That level of governance and human-in-the-loop refinement creates a far better feedback loop than external tools allow.
|
Layer |
Technology |
Purpose in our Culture |
|
The Foundation |
Snowflake |
Perfect for the tech team. We are leaning in heavily here to build a robust data infrastructure. |
|
The Interface |
ThoughtSpot (Spotter) |
The essential bridge. It makes complex data instantly accessible to the wider organization without requiring users to navigate the underlying infrastructure. |
The essential bridge. It makes complex data instantly accessible to the wider organization without requiring users to navigate the underlying infrastructure.
3. Elevating the Data Analyst Up the Value Chain
Let’s be clear about one thing: AI self-service analytics does not magically build its own foundations. Someone still needs to build the trustable context, model the data correctly, and manage the infrastructure.
The secret for us has been clearly defining what work is self-service versus what is analyst-led. Instead of spending their days answering simple, repetitive data questions, our analysts are moving up the value chain to focus on highly complex, strategic problems.
💡 Our New Reality
"Our data team are like consultants for our business now. They can work on things that drive growth for the company rather than simpler data questions which the business can now get themselves."
To give you an example of what this looks like in practice: Our CEO recently dropped a complex, ad-hoc business question into Slack. Because our data foundations were solid and our team was freed up from basic reporting, we were able to leverage our tools to return a fast, comprehensive executive summary in a matter of minutes—not days.
My Advice for Getting Started
If you are trying to increase AI adoption in a way that your organization actually trusts, here is my personal playbook:
Give people the freedom to "break the system": Don't lock data down so tight that people are afraid to touch it. Ask people to see how easy it is, and give them the liberty to build their own visualizations. Even non-technical staff can become empowered to edit models once they see how simple the process actually is.
Build trustable, contextual data models together: Scale your data culture with your business users, not in a data team silo. Do it together. Bring them in to help embed the "why" behind the numbers into your data models as you scale.
Stop planning, start building: Data people are, at our core, builders. This is the ultimate era for us to get our hands dirty and build. The technology is here, and it has never been easier to make a massive impact.
What does the next step of your AI roadmap look like? Let’s connect on LinkedIn to keep the conversation going, or reach out to the team at ThoughtSpot to see how they are making agentic analytics a reality.
