This week, we had a chat with one of our enterprise customers about their thoughts on AI inside their Yellowfin environment.
They’re currently running version 9.15 and use Yellowfin as the primary BI layer inside their app.
Upgrading to 9.16 for them will be a whole different experience – as of that release, we really explored how well we could make AI benefit Yellowfin in a way that strengthened its existing features and safely added capabilities which take things further in a positive direction.
We’ve made a range of improvements to our AI NLQ. If you are not familiar with our intelligent Natural Language Query (AI NLQ), then I strongly recommend reading this article as it covers it in depth.
One theme that seems to have really caught people’s attention is that data exploration rarely happens in a single step.
It’s not an old-fashioned simplistic query:
“Show me revenue by region.”
It’s a conversation, where you keep going (and Yellowfin understands and remembers the context), and refine what you want, based on the results you’re shown:
“Show me revenue by region.”
“Okay, now exclude Europe.”
“Now compare that to last year.”
“Now break that down by product line.”
More like a human colleague, helping you to get the answers you need.
We also talked about the friction that can arise when some of their users turn to external AI tools when they need help interpreting data.
They’ll open a browser, use Copilot or another AI chatbot to make sense of a dataset, and then return to Yellowfin to continue their analysis.
There’s nothing bad with this approach, as such, but we want to make things seamless and efficient for end users, and that’s why we have expanded our AI model support to work with the top AI LLM providers.
At Yellowfin we’re not locking you to one specific AI platform, like some other larger BI vendors do, we’re offering choice. You pick what works best for you, your budget, and your particular business circumstances.
Another detail that people seemed to love was editability. In Yellowfin, AI-generated charts can be adjusted like any other chart in Yellowfin. They’re not fixed outputs. Users can refine them, incorporate them into Yellowfin’s excellent Stories or Present, and shape them further.
That balance between automation and human control makes AI practical BUT it’s not without its caveats. We all know if you throw caution to the wind AI can “hallucinate” – quite literally make things up – or potentially leak sensitive information which is why we’re very careful with what we send as inputs (never row-level data). We are working for, and with, you – and that’s a partnership we take very seriously.
“Our continued investment in AI functionality demonstrates Yellowfin’s commitment to developing products that deliver tangible benefits to our customers,”
– Ethan Brown, Yellowfin GM.