Getty Images/iStockphoto
ThoughtSpot AI agent Spotter enables conversational BI
The analytics vendor's new tool not only enables users to converse with data using natural language, but also has contextual awareness and autonomously makes suggestions.
ThoughtSpot on Wednesday launched Spotter, a generative AI-powered agent that enables customers to have conversational interactions with their data.
The vendor first unveiled Sage, a generative AI-powered interface that let users query and analyze data, in private preview in March 2023, making it one of the first analytics vendors to introduce a feature integrating large language model (LLM) technology with its existing AI-powered search capabilities. Sage, however, was never made generally available despite later moving into public preview and adding capabilities during the preview process.
Spotter now replaces Sage, according to Sumeet Arora, ThoughtSpot's chief development officer, with Spotter for ThoughtSpot Analytics now generally available and Spotter for ThoughtSpot Embedded in beta testing.
Unlike early generative AI-powered assistants such as Sage and copilots from Microsoft that enabled users to ask questions of their data using natural language and receive answers, agents go beyond mere question-and-answer capabilities to become more autonomous. Spotter and other agents know context, enabling them to do more than a chatbot without being prompted.
They can suggest follow-up questions to initial queries to enable deeper analysis, surface insights that might not otherwise have been discovered, monitor for anomalies in data and make suggestions. Agentic AI is more proactive than the reactive AI tools many vendors developed before agents emerged in recent months.
Donald FarmerFounder and principal, TreeHive Strategy
As a result of the analysis it enables, Spotter is a significant new tool for ThoughtSpot users, according to Donald Farmer, founder and principal of TreeHive Strategy.
"To my mind, Spotter is a big move forward for ThoughtSpot and AI," he said. "The natural language interface is improved, so it's much more conversational rather than just search-based. But the key advantage will be the autonomous analysis, which can identify trends and insights without requiring users to specifically ask for them."
Based in Mountain View, Calif., ThoughtSpot is an analytics vendor whose search-based platform enables users to query and analyze data.
In September, after Sudheesh Nair's resignation in March, former Salesforce executive Ketan Karkhanis was appointed CEO.
New capabilities
ThoughtSpot's platform has always been centered around natural language processing (NLP) to try to make BI accessible to nontechnical users.
The use of analytics tools to inform decisions has long been limited to a small percentage of trained users, often data experts, due to the coding skills typically required to prepare and analyze data. NLP aimed to reduce the technical skills needed to work with data. However, the limited vocabularies of NLP tools, including ThoughtSpot's platform, meant users still needed training before using even self-service analytics platforms.
Generative AI changes that.
Large language models such as ChatGPT and Google Gemini have vocabularies as extensive as any dictionary, thus enabling true free-form NLP. When integrated with an organization's proprietary data, they enable users of all skill levels to ask questions of data and make data-informed decisions.
As a result, many analytics and data management vendors have spent the past two years -- since the November 2022 launch of ChatGPT marked a significant advance in generative AI capabilities -- developing generative AI tools that customers can combine with their own data.
Sage was ThoughtSpot's initial foray into generative AI. Spotter now delivers not only a question-and-answer interface, but also reasoning capabilities that enable deeper analysis than is possible with a single query -- essentially serving as an AI-powered analyst, according to Arora.
The tool, however, is limited to structured data analysis and cannot analyze unstructured data such as text, images and audio files.
"Spotter enables our customers and users to interact with their enterprise structured data in a conversational and iterative humanlike way, delivering precise, interactive answers to any query, regardless of where the data resides or its complexity," Arora said. "We are inspired by and are building the skills of an analyst into Spotter."
Specific features of Spotter include the following:
- Natural language question-and-answer capabilities for structured data wherever the data lives, no matter the size or complexity of the data set.
- Contextual understanding of a conversation so that users can ask follow-up questions after an initial query, with Spotter remembering the initial query as it responds to additional questions.
- Continuous learning so that Spotter gains greater understanding of an enterprise's language and operations to provide more tailored insights based on human feedback.
- Extensibility beyond the ThoughtSpot environment, with Spotter able to be embedded into business applications such as Salesforce and ServiceNow so that users don't have to lose time toggling between environments to get data-driven insights.
- Security and governance including role-based access controls so that users can only access data they are authorized to use.
- Compatibility with any cloud platform and popular LLMs, including ChatGPT and Google Gemini.
Mike Leone, an analyst at TechTarget's Enterprise Strategy Group, said that with its AI agent, ThoughtSpot is meeting users' needs.
About two-thirds of organizations are either planning to build AI agents or considering developing them, according to Enterprise Strategy Group's research. Spotter provides ThoughtSpot's customers with the means to do so.
"Having an AI agent story is critical right now," Leone said. "Spotter gives ThoughtSpot an early agentic message for customers."
Particularly significant is Spotter's ability to learn about its users and their organizations to gain context and deliver personalized responses, he continued.
"An area that is so important as organizations pursue GenAI initiatives is the ability to understand nuance," Leone said. "One of those nuances is the type of person that is engaging with the underlying system. Spotter is able to adapt ... to deliver personalized and contextually relevant responses, which goes a long way in delivering unique value across a business."
Farmer, meanwhile, highlighted Spotter's embedded capabilities, noting that ThoughtSpot's embedded analytics tools have gained momentum and become an important part of the vendor's platform.
"The embedded cross-platform integration is important," he said. "I am seeing ThoughtSpot gaining a lot of attention as an embedded analytics solution, and Spotter's integration with business productivity tools and its ability to drive custom agents should be very attractive."
From a comparative perspective, Spotter brings ThoughtSpot's agentic AI capabilities in line with those from other vendors, Farmer continued.
Google is one big vendor taking an agentic approach to analytics, introducing Conversational Analytics in Looker in September. Salesforce is another, with Tableau Agent now part of Salesforce subsidiary Tableau's platform.
"In general, ThoughtSpot is comparing well with other vendors who are integrating AI capabilities to business intelligence tools," Farmer said. "Each [vendor] is taking a unique path, and ThoughtSpot's route is proving to be fruitful: building on their existing specialty of search-based analytics. In general, ThoughtSpot is integrating AI in an interesting and effective way."
Leone similarly said Spotter's launch is in line with the agentic AI tools being developed by competing vendors. Like others, ThoughtSpot is starting with key capabilities -- in its case, search -- with plans to expand its agentic AI functionality more broadly over time.
"ThoughtSpot is right in line with the market in terms of delivering an agentic experience within its platform," Leone said.
Next steps
Although Spotter is now generally available in ThoughtSpot Analytics, the vendor has plans to continue improving the AI agent's functionality, Arora said.
Currently, the tool mimics the skills of a trained analyst. Soon, it will be able to proactively tell users about their data, suggest questions for users to ask of their data and answer questions about why something happened.
Ultimately, the goal is for Spotter to be an autonomous AI agent, according to Arora.
"Eventually, it will be able to take action on your behalf as well," he said. "This is what we mean by autonomous."
Farmer, meanwhile, suggested that one way for ThoughtSpot to expand and improve its platform would be to add industry-specific applications.
Many vendors, including Databricks, SAP, SAS and Snowflake, offer prebuilt industry-specific applications to make it easier for new customers to get started with their tools and increase speed to insight for existing customers.
In addition, ThoughtSpot could make its generative AI-powered analysis multimodal, according to Farmer.
"A natural extension to natural language would be multimodal capabilities [such as] speech interfaces and access to video and audio," he said.
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.