As an industry analyst, I first identified ThoughtSpot as a leader in AI-driven Business Intelligence back in 2019. At that point, they were already using AI to provide users with assisted insight, automated insight, natural language explanation, next best insight, natural language queries, voice recognition, visualization recommendations, image recognition, next best question, and autofill search recommendations. Much of their AI leadership, at the time, was due to their unique semantic layer and active use of metadata. With their release of ThoughtSpot Sage in March 2023 and their recent 31 January 2024 announcement, ThoughtSpot continues to accelerate their leadership position in the use of AI in Business Intelligence.
According to Sumeet Arora, Chief Development Officer at ThoughtSpot, “…we augmented our patented relational search technology with the power of LLMs to provide a natural language front end on top of a robust, battle-tested computational engine that delivers accurate results on complex data at scale.”
The Data Renaissance
As someone who lives just 40 minutes from Florence, Italy, I find it fitting that ThoughtSpot sees generative AI as a technology that is ushering in a “data renaissance.” The idea is that “anyone can ask questions like a business leader and get answers like an analyst.” In this recent announcement, ThoughtSpot launched the following capabilities:
Advanced human-in-the-loop feedback. To help fine tune generative AI outputs, a patent-pending feedback loop enables users to teach and train ThoughtSpot Sage by providing associations between business terms and phrases and a data model. In addition, users can view and edit all queries and generate word and phrase-specific feedback. A human feedback loop is essential to the continual improvement of accuracy in the use of AI.
Conversational BI. Ask Sage lets users converse with ThoughtSpot Sage from within the context of each data visualization, using natural language to ask their questions. Users can ask followup questions, drill down for more detail, and see the SQL behind their answers, all within the context of their original search for insight. The use of conversational interaction with AI opens the door for data literacy and the expanded use of insight by a non-technical audience.
AI assistance. AI Assist takes generative AI to Mode where users find help generating and improving SQL. Users can use natural language to refine, create, understand, troubleshoot, and optimize queries. AI Assist is unique in the fact that it seamlessly allows users to move between SQL and natural language. Samantha Novak, Product Design Lead at Mode, instructed potential users, “Lean on SQL keywords and valid syntax when it feels more natural, switch to English shorthand when it feels more natural.”
AI highlights on Liveboards. When users return to their desktop after a time away, AI Highlights extracts important takeaways from a Liveboard, including both expected and unexpected changes and the attributes driving the most change. With personalized, automated insights, analysts and business users are much more likely to find actionable insight quickly.
KPI Monitor. Instead of waiting for users to login and check on a metric, KPI Monitor sends alerts for what’s changing in the business and why. The alert includes a generative AI, natural language explanation that is delivered through the ThoughtSpot mobile app. This move toward monitoring and alerts addresses the static, non-realtime nature of dashboards and enables mobile users to make just-in-time decisions from remote and global locations.
Embedded natural language search. Generative AI is also making its way to ThoughtSpot Everywhere with the introduction of Sage Embed. Users can now bring ThoughtSpot Sage into every analytical product and web app built on the ThoughtSpot platform.
ThoughtSpot customer Act-On is a marketing automation company driven to provide marketers with real-time insights that help them make quick and efficient strategic decisions. After installing Sage Embed, the company claims to have seen a 60% increase in customer report usage, compared to old reports.
AskDocs. With an AI engine at the core of its platform, ThoughtSpot also expanded its use of generative AI by applying it to their product documentation. With AskDocs, developers can get up to speed and build new applications much more quickly by asking product and coding questions of the documentation, instead of searching for answers. Developers can ask questions in natural language and receive instructions and code from the Visual Embed SDK in both ThoughtSpot and developer documentation.
“To build ThoughtSpot Sage….we augmented our patented relational search technology with the power of LLMs to provide a natural language front end on top of a robust, battle-tested computational engine that delivers accurate results on complex data at scale,” said Arora, describing his take on what is different about ThoughtSpot. “Transparency is available at every step as well as advanced human-in-the-loop verification and training controls for users, analysts, and administrators. And we are thrilled to share that customers have reported up to 95% accuracy using ThoughtSpot Sage when compared to GPT alone.”
Ferraro Consulting POV
Back in 2019, I predicted that business intelligence vendors who were investing in the use of AI to automate formerly manual tasks and make insight available to more users would be the winners. The explosion of generative AI solidifies this assertion. Vendors, like ThoughtSpot who built their platform with an AI engine and natural language capabilities, take the lead also in the use of generative AI. Ferraro Consulting expects ThoughtSpot to continue to expand their lead in the use of AI.
In addition, generative AI lacks the guardrails necessary to deliver accurate and reliable analytical results. Ferraro Consulting believes that the responsible use of generative AI must include a rich set of metadata, a common language business model, a human feedback loop, and a validation loop in order to produce results on which decisions can be made. ThoughtSpot is delivering significant technological advances in all of these areas. It is important for companies making decisions about business intelligence platforms to dig below the surface of generative AI “tack-on” demos, and to understand how vendors are using metadata, common language, and feedback to fine tune their outputs.
© Ferraro Consulting, 2024