ThoughtSpot users can now embed its analytics into their own customer-facing applications

Business intelligence firm ThoughtSpot Inc. today announced a fresh batch of features aimed at pushing analytics beyond dashboards and into customers’ applications and workflows.

The rollout includes a new version of its Spotter agent, dubbed Spotter 3, and a set of enhancements to its embedded analytics product that enable customers to expose analytics in their own applications.

ThoughtSpot said it aims to treat artificial intelligence not as a co-pilot or interface layer but as a foundational agent that can be embedded, reason through decisions, and adapt to different user needs.

“The whole idea is breaking the boundaries,” said Francois Lopitaux, ThoughtSpot’s senior vice president of product management.

The embedded offering is designed to make it easier for companies to offer analytics to customers within their own applications, eliminating the need to build the full stack from scratch. That includes exposing the Spotter agent directly via a software development kit and application programming interfaces.

“Our customers now can embed Spotter into their own applications for their customers, which was not possible before,” Lopitaux said. Embedding also enables customers to monetize the analytic services.

He cited Navan Inc., maker of enterprise-oriented travel management software. The firm recently unveiled analytics features powered by ThoughtSpot that distill over 100 complex travel data points — such as departmental spending, out-of-policy bookings and savings — into interactive visualizations, freeing users from the need for spreadsheets.

“Navan is exposing analytics to me as the customer,” Lopitaux. “Our CEO can easily see spending money by month, without going over the target.”

Have it your way

Customization is enabled by ThoughtSpot’s Muze Studio for visualization and Dev Studio for SDK-based customization. “If I take somebody else’s software and put it inside my product, I want this to look like my product,” Lopitaux said.

The studios allow developers to customize style sheets, hide buttons, subscribe to events and build workflows that trigger actions in tools like Salesforce Inc.’s customer relationship management system and Slack collaboration software.

The platform integrates with existing agents using a Model Context Protocol server, allowing customers to plug Spotter into their own agent systems. “We allow our Spotter agent to be integrated within their own agent,” Lopitaux said. “They can unify them into one unique agent.”

Spotter 3 builds on the previous generation of the company’s natural language agent with deeper integrations of structured and unstructured data, new skills like forecasting and support for Python coding and the ability to explain its reasoning.

“If you ask how many jackets you sold in California last week and the answer is zero, Spotter3 is going to say, ‘Why is it zero?’” Lopitaux said.

The agent can now forecast trends, detect anomalies, write code and perform basic machine learning tasks, such as bucketing, a machine learning technique that clusters elements similar to each other.

Spotter connects to data sources like Snowflake, Databricks and Google BigQuery as well as business applications like Microsoft Teams and Slack. The software features some data transformation capabilities, but it isn’t a comprehensive integration package.

Spotter 3 will start rolling out in November. The embedded features are generally available beginning this week. They will carry an extra charge based on usage. ThoughtSpot publishes its prices on its website.

Image: Tung Nguyen/Pixabay