Amazon Web Services Inc. said today it’s expanding the generative artificial intelligence capabilities of its cloud-based business intelligence service QuickSight with agentic AI.

Amazon QuickSight is a platform for business users that provides users with machine learning capabilities that can reach into company data and provide spreadsheets, charts and visualizations.

It includes a generative AI capability integrated with Amazon Q, an AI-powered assistant the company unveiled during re:Invent 2023. It can answer questions, provide summaries of data and complete simple tasks based on natural language queries such as generating dashboards.

Today, Amazon released “Scenarios,” a new way for business users to approach their data by allowing them to become a data analyst. Previously Q in QuickSight opened up opportunities for authors — the creators who built pages and visuals — and consumers, the users who need summaries and questions answered.

The new offering opens up the opportunity to make every user an analyst who can dive deeply into data and go hands-on with spreadsheets and datasets. This capability was originally demonstrated during re:Invent last year in preview and is now generally available.

“We know that every user doesn’t want to go in and understand how to do regressions and complicated analysis,” Jose Kunnackal, director of Amazon QuickSight at AWS, told SiliconANGLE in an interview. “But we do know that users at some point might use a spreadsheet to manipulate data to get an outcome. Scenarios in QuickSight allows a regular business user to do this by simply stating a question and in the background it does the analysis for them.”

In a normal interaction an AI assistant, a user might ask a question and have it generate a visual, such as a chart, a graph or a dashboard. Then have it refine the result by asking another question or iterating conversationally. According to Kunnackal, this method is very slow and plodding.

Scenarios accelerates analysis through agentic AI, which automates complex, multistep data analysis. This means that a user can simply type in a scenario in conversational English and an AI agent will seek out data, formulate a plan and execute the necessary steps, significantly reducing effort and time on the user’s part.

Kunnackal highlighted that the AI agent shows its analysis process so the user can follow along and derive the path that it took. Although the AI agent is doing the deep analysis work, it’s not a complete black box. This allows users to refine their approach by seeing where the agent went and how it chose to go there.

For example, a user could provide “what-if” scenarios to the AI agent and it could look through proprietary company data or uploaded documents and then take a few minutes to do an analysis. These are tasks that would take a data scientist or analyst weeks or days – now reduced to minutes.

Each time a user asks a question the AI agent produces a thread where the user can add follow-ups. Users can also explore multiple related or alternative inquiries from an initial question forming simultaneous threads. That allows users to avoid linear thinking in data exploration and investigate multiple paths at once, allowing for more creative outcomes, Kunnackal said.

For example, a user could examine product sales performance between different regions across separate threads, the impact of price changes on revenue and potential inventory adjustments. That approach helps mimic the natural human brainstorming process of exploring many ideas at once but lets the AI do the complex analysis and deliver a clear presentation.

“Scenarios is a natural progression of the work we’ve been doing in QuickSight,” Kunnackal added.

With this addition to Q in QuickSight, Kunnackal said AWS is looking to broaden deep analysis capability to more users. After unveiling the capability last year, it has been in the hands of early customers such as BMW Group and Amazon itself where it has been put to use to uncover insights in supply chain data and to analyze support ticket patterns.

Image: Amazon Web Services