In the age of generative artificial intelligence (AI), everything is beginning to be reimagined. It’s hard to escape the buzz around this new technology that holds the potential to bring in constructive changes into every possible industry.
Business intelligence, at its core, drives good business decisions. It helps make sense of a lot of disparate datasets. And given what we’ve seen from the early successes of generative AI with ChatGPT, BI seems like the perfect use case that will benefit from this technology.
To understand the scope of what this means and what possible ways BI could change, we spoke to Benn Stancil, chief technology officer and co-founder at Mode, a business analytics and business intelligence platform. Benn is known for his craft, as well as his blog, where he writes candidly about his experiences as a person of the technology industry.
In the fourth episode of Dataverse, Manjot Pahwa and Ravi Suhag discussed the present and future of BI, the modern data stack, and the impact of generative AI on the data industry.
Consolidation in the modern data stack
The modern data stack, simply put, is a set of software tools that help collect, process and store data in order to derive insights. Organisations value this because it saves them time and helps make better business decisions. Over time, the number of these tools that simplify data-related processes has grown. A question that often gets asked about the future of the modern data stack is about consolidation.
In Benn’s view, a consolidation can happen in many forms, and is good for customers. It could be through acquisitions of companies, or customers consolidating the kinds of tools that they want to use. A few years ago, when the concept was new, several companies were able to raise funds and find paying customers relatively easily. When the market changed, fundraising slowed down, and customers began realising they didn’t want to pay $50,000 for every data tool, things began rationalising.
“Fundamentally, how we consume data is mostly a point of preference and personal ability,” Benn remarks. So it’s equally likely to find customers who want to interact with data on spreadsheets, and some others who prefer SQL or Python.
Ravi feels consolidation can happen at two levels: first, in the data governance layer, where discovery, observability, data contracts, data tools and data access begin to form an experience, and second, in the ingestion transformation lifecycle.
Another layer of consolidation, Ravi feels, needs to come in at the level of defining a metric that analytics tools rely on. For instance, there should be a uniform meaning for active users within an organisation in order for the data practitioners to make good inferences.
Advantage generative AI
With every tool promising great insights or business decisions, where do things really stand? Is data actually helping companies progress faster?
Ravi explains the existing data scenario for businesses with an interesting analogy. Think of a business as an automobile. The various data tools are like metres that have the ability to measure engine performance, fuel level, speed, and so on. There are different people who have access to these metres, but not all are able to figure out why the automobile is behaving a certain way. Real value will come from being able to read the data and say why the automobile is moving at a certain speed, where it may be experiencing an issue, and how to make it operate at full capacity.
“There are a very small percentage of businesses where data is actually converted into actionable business decisions. That is where the gap is,” he says.
With generative AI, there is an opportunity to fill this gap, says Benn. The key differentiator with this model of AI is its natural language processing ability. And it has the potential to be disruptive because it essentially automates what an analyst does.
At the end of the day, AI has the ability to make data teams smarter in the same way that Google makes us all smarter, he adds. Where things change is access. “Say I am a marketer, and I understand the business much better than an analyst does. I can translate a question from the market into something that actually the data can inform,” he says. Essentially, business functions across the board will be able to glean data insights to drive better business outcomes.
Evolving roles
Chatbots and the power of AI are themes that have been part of technology discourse for almost a decade now. The big difference with the generative AI wave, says Ravi, is natural language processing capability through voice, and not just text or chat. “It reduces the learning curve for any tool within the modern data set,” he adds.
For Benn, it is also the inherently interactive nature of this new technology that is exciting. A product such as ChatGPT, for example, will ask for feedback and clarification. “It can ask if I want to see revenue by week and it can clarify, do you mean ARR? Or do you mean bookings? It can follow up with you, and that, to me, is a huge difference. That’s why I think it actually can start to replicate what analysts do because it’s not just attempting to take a question and say here’s our best guess at the answer. If it’s not sure,it will say we don’t know what to do,” he adds.
So will analysts cease to exist? Benn disagrees with the common narrative that AI will do the rote tasks, leaving humans with the time to do creative things. “But ChatGPT is good at creative stuff. You give it a weird business problem and say, give me some hypotheses as to why you think this might have happened. And it will do a pretty good job. So there will be some pressure on analysts,” he says.
But what will be valued is general awareness of everything that’s going on in a given sector. The little insights that qualify a person’s understanding and experience. Analysts’ skills, Benn feels, will evolve to cater to these new requirements.
Changing contours of the data stack
Does this mean that the modern data stack will see a complete overhaul with AI integration, or will incumbents have an advantage in building the next set of data tools compared to startups?
Benn explains this with an automobile analogy of a different sort. He compares the AI advancement of the modern data stack to a self-driving car. A car manufacturer has the car, so is naturally better positioned to integrate self-driving capability. But years of knowing car-related stuff will have to be reimagined. A completely new way of driving, speed and other parameters will have to be understood. Building a ChatGPT-enabled BI tool will be a similar exercise.
Existing BI tool companies may have an edge at the start, but it won’t be an easy task to automate everything. Going back to his point on customers having their own preferences in consuming data, Benn points out that the journey will not be a simple plug-and-play solution. Several startups are toying with new models, one of which is building Slackbot-type capabilities.
Using generative AI for root cause analysis might be a more effective way of ChatGPT integration, he adds.
One thing in the data stack Benn feels will undergo a significant change is documentation. Over time, given that data insights and entire dashboards can be generated on the fly, having static documentation could become redundant.
For Ravi, the one change in data and analytics tools could be the way they interface with customers.
“Whether in chat or through voice, tools backed by AI will start to become headless, or provide very little interface of their own, and they will start to just directly integrate with systems,” he says.
Looking out to the future
Given the fast moving pace of change being brought in by AI, where do data practitioners see a radical change in the future?
For Benn, it will change the way of looking at data. A lot of data dashboards were supposed to make everyone smarter, but so far, that hasn’t happened. He doesn’t offer an alternative, but thinks there could be different ways to think about the utility of data that is quite literally at people’s fingertips.
Ravi has another interesting take: “I think our use of fingers will change. If we look at human evolution, we have literally made our fingers do something which they were not meant to know, which is type on computers and interact with machines. And I think interaction with voice is a more natural interface to do that,” he adds.
Irrespective of how data and its practice evolves, we are witnessing a technology that has the power to transform the world, one industry at a time. Our job as data practitioners is to ensure that this change brings a more meaningful transformation for businesses.
If you too are a data practitioner who would like to be a part of Dataverse, please sign up here or contact me at manjot@lsip.com.
Authors
