Like bell-bottom jeans in the '60s, pastel colors in the '80s and cargo pants in the '00s, generative AI is all the rage in the 2020s.

But unlike fashion fads, while it is perfectly possible that Jeff Bezos wore getups that went with the times, AWS is all in on generative AI, which includes infusing its QuickSight analytics platform with generative AI capabilities.

Generative AI has the potential to be as transformative a technology for enterprises as perhaps even the computer itself by making workers exponentially more efficient. At the core of generative AI is data, which is what provides generative AI tools with their intelligence.

As a result, many data management and analytics vendors have made generative AI a focal point of their product development since OpenAI's November 2022 launch of ChatGPT marked a significant improvement in generative AI capabilities and spurred the surging interest in the technology.

Tech giants such as Google, Microsoft and Oracle have all made generative AI an important part of their respective data platforms. So have rising data platform vendors Databricks and Snowflake, as well as more specialized data management and analytics vendors such as Alteryx, Informatica and Tableau.

AWS is no different. Generative AI now dominates its database strategy. And it is the dominant theme of AWS' analytics strategy as well, with QuickSight now offering Q, a suite of generative AI capabilities. Q includes tools that enable customers to use natural language to query and analyze data, develop narrative summaries, and even create dashboards and other data assets.

What once required copious amounts of coding that took even trained experts significant amounts of time can now be done in minutes.

With AWS continuing to expand its generative AI capabilities in QuickSight, Tracy Daugherty, QuickSight's general manager, recently discussed the platform's inclusion of generative AI tools and plans to develop additional ones.

In addition, he spoke about the rate of adoption of QuickSight's generative AI tools, what types of enterprises are among the early users and what is holding others back, and whether QuickSight -- which is tightly aligned with other AWS products -- is gaining traction with customers who don't use AWS for their complete data stack.

Editor's note: This Q&A has been edited for clarity and conciseness.

How Generative AI Is Changing Data Engagement

How are you seeing generative AI change the way businesses engage with data?

Tracy Daugherty: I see it on two different avenues. The first one -- and we're doing this with QuickSight with authoring capabilities and calculation capabilities -- is to make it easier to do something that was either time-consuming or difficult. Think of it as having a job that has to be done, but is now either more approachable to more people or the time is cut down in a dramatic way for more advanced people. That's one half.

The other half is enabling people to do something they couldn't have done before. When we added multivisual Q&A capabilities, it allowed the everyday business user to ask questions of their data that went beyond what was in a report or dashboard.

The trick is that most business users don't know how to ask the question and often aren't even sure what question to ask. The experience walks them through and gives them multiple options. In the end, people wind up telling us, 'Man, you gave me the real answer I was looking for, but I didn't even know that was what I wanted.' They got an insight that wasn't just an answer to their initial question, but was instead something else impactful.

Comparing Generative AI With Traditional Methods

How is that different from what analytics consumers could do before?

Daugherty: In the history of BI tools, it was someone talking to someone on the back end and saying they want a report to look a certain way. Then they'd get the report and say they want it to look a little different. It became this back-and-forth, which stems back to the idea that users often don't know what they're looking for until they see it. With GenAI, we're broaching this new way for them to get insights without the involvement of others in an insightful and powerful way, which is exciting.

Demonstrating Generative AI's Use

Can you give an example of how QuickSight is making it easier to do things and enabling business users to do things they might not have been able to do before?

Daugherty: We've launched a feature called Data Stories where you can create a story from your data. Most people, when they explain their problem to someone, do it in a PowerPoint or in a document. They don't point to a dashboard and say, 'Here, you figure it out.' GenAI has helped QuickSight users to select a set of data to do a visual representation and then generate the writing that comes along with it to make it shareable. Could they have done that themselves? Sort of. They could have taken a screenshot and written things up, but that's hard work, and people often wind up deciding it's too hard, and it never gets done. Now, if GenAI gets them 90% there, and all they need to do are a few tweaks, it becomes another example of saving time and doing something that otherwise would have been too hard.

If I were to net that out, GenAI allows a business user to be more data-driven in their decision-making and sharing their thoughts with others. That's the biggest benefit.

QuickSight's Specific Generative AI Capabilities

What are some specific generative AI capabilities that are now part of QuickSight?

Daugherty: For the person who's an analyst who builds dashboards and reports, now through natural language they can ask a question, and QuickSight prebuilds the visual, connects it to the data, and the user can edit the visual by using natural language. For example, they can tell QuickSight to make it a bar chart and then add filters. It reduces the overload of having to be an expert about where the data sits to take action and greatly speeds up the creation of dashboards.

On the end-user side, it's the Q&A things I talked about, the multivisual Q&A and the Data Stories. The multivisual Q&A is not just asking a question and getting an answer. That doesn't work for querying databases and data warehouses. You have to get an answer that's both visual and text-based, so we provide that experience. Then we provide an experience that asks whether the user meant something a little different in case what they got back wasn't exactly what they wanted. Data Stories is just empowering end users to go beyond their dashboards and reports to build a story to go with their data. If you were to ask me how my business is doing, I wouldn't send you a dashboard. I would have some points and build a presentation with a story, and now that can be done through generative AI.

A Use Case of QuickSight's Generative AI

Can you describe the use of one of QuickSight's generative AI features?

Daugherty: What I'll talk through is the authoring experience if a user were to build a dashboard visual by visual.

Let's say you start by asking for sales by location as a map. QuickSight will wire that to the data in the system. Then you can click a button that says 'Add to Dashboard,' and the visual map gets added. Now you can do another one such as forecasting profits by month. QuickSight will automatically choose what graph type to use to display the data, but the user can select a different graph type if something else is preferable. Then you can add calculations, which are super complex and usually require SQL. You can use natural language, and it becomes a calculation you can use in your dashboard.

In a matter of 60 seconds, you can create a dashboard with two visuals and a calculation.

How Time Is Saved

How long might that have taken without generative AI?

Daugherty: If you were an experienced user and knew all the data and all the tools, it would have probably taken 45 minutes. If you were new to the tool, it would have taken a day or two because you wouldn't have known what data to wire the dashboard elements to. Picking the chart and that type of stuff is easy. The hard part is knowing where the data is.

When you add up that time savings, it becomes pretty powerful.

Q in QuickSight

Are these generative AI capabilities all components of Q in QuickSight or do they have their own place in the platform?

Daugherty: These are all what we put under the umbrella of Q in QuickSight. The way I describe it is that Q grew up. Initially, it was just Q&A, but now it has expanded.

Adoption and Hesitation

Are you seeing widespread adoption of QuickSight's generative AI capabilities or are you finding that many users are holding back until generative AI matures a bit further?

Daugherty: I don't have specific numbers, but since April 30 when Q was made generally available, we've had a big uptick in the adoption.

We learned that people want to feel that it's safe and secure and accurate. Our authors building dashboards and reports find it highly valuable, highly effective and highly safe. It just helps them do a job. It's not magic. It's doing something applicable that they know and helps them save time. The multivisual Q&A experience has also made people feel safe because we're not tying responses to being 100% accurate. What we're doing instead is enabling users to ask some generic questions and providing generic responses so users can then dive deeper to get insight. That has increased confidence.

Data Stories is a new thing. What we find is that there's a subset of people that get super excited and use new features right away, while others want to see how to use it first and figure it out.

Early Adopters vs. Late Adopters

Are there common traits you see among the early adopters of QuickSight's generative AI capabilities?

Daugherty: There's not a common trait as far as industry, but there is a common trait as far as mentality.

The ones that tend to be more aggressive tend to have leaders who believe generative AI is greatly going to help them. They tend to have a view that they're a tech company first, even if they're in an industry you don't think of as being a tech company. They have the attitude that they want to lead -- generative AI is a competitive advantage, and they want to invest in it. Those are the ones we see connect earlier.

Conversely, are there common traits you see among those not yet adopting QuickSight's generative AI capabilities?

Daugherty: Like anything in the world, you have another set of companies that are more conservative. They want to see examples of other companies that generative AI has worked for and how it's worked for them. They don't want to go through the growing pains and want to get to the end solution right away.

QuickSight's Future with Generative AI

As generative AI continues to evolve, what more can an analytics platform such as QuickSight provide to customers?

Daugherty: In short, a lot.

I won't speak to the developer side because AWS has other folks who are more ingrained in that area. From the business user and analytics side, we've only scratched the surface of what we can do. Most of the work until now has to do with reporting and making people more aware of data. The next step is to become more proactive about what's happening and letting people know. There's a lot we can do there. There are things related to data preparation that we can add. I put a lot of energy into figuring out how the end user can get more value from their data. This is an area where we're constantly pushing, but we're also learning a lot in a short amount of time. The trick is to provide instant value that people can trust, and then their organization has to let the people use it so that they can also trust it.

There's an interesting challenge because business users aren't very forgiving. If they try something that doesn't work, then they tend to not go back to it. So, there's a lot of pressure -- but in a good way -- in terms of nailing the experience, science and quality.

Other Recent Features of QuickSight

Beyond generative AI, what are some other features added to QuickSight in recent months?

Daugherty: The one thing that isn't related to GenAI is that we introduced pixel-perfect reporting.

We now have both dashboarding and pixel-perfect reporting. Everyone knows what a dashboard is. The analogy I give for the difference between a dashboard and a pixel-perfect report is that when you go to your power company and look at your usage, that's a dashboard, but when you get your itemized bill, that's a pixel-perfect report. It's a very formatted, structured thing. That is gaining a lot of traction.

Another thing we introduced is added pricing options. We used to have two types, a reader and an author. We added a reader pro and an author pro with elevated capabilities. The pro users get all of our GenAI capabilities included, while the multivisual Q&A feature is available to all four user types.

QuickSight's Deployment Models

Given that QuickSight is an AWS platform, it obviously integrates well with AWS data management tools, but do you see many customers develop data stacks with QuickSight in concert with data management tools from other vendors?

Daugherty: Obviously, it's a first choice for AWS customers because it's integrated. The second-most common way we see QuickSight deployed is virtual private cloud connections to on-premises solutions. For example, they may have Microsoft SQL Server on premises or another preferred database on premises. And typically, in that instance, it will be part of a migration phase. They'll use QuickSight with their on-premises database, and then move that other part to the cloud.

We don't see as many cross-cloud deployments -- we do see it -- for a number of different reasons, but usually we're seeing customers try to get from on-premises to the cloud, and then build their story from there.

QuickSight's Roadmap

Lastly, what can you share about QuickSight's roadmap, both in general and as it relates to generative AI?

Daugherty: In BI tools, we're historically very tied to structured data. As you can imagine, there's more data out there that we want to bring in to make QuickSight more comprehensive. We're doing a lot of thinking about how to do that.

It goes back to the idea of making the business user's life better, stronger and more complete. We have to think about all the data they can have access to in order to make it more complete. It just so happens that all the structured data in databases and data warehouses is very black-boxed in a user environment we're experts at. But there's other data we can combine with that to make the data even stronger.

Eric Avidon is a senior news writer for TechTarget Editorial and a journalist with more than 25 years of experience. He covers analytics and data management.