Just like cloud computing changed digital businesses over time, the momentum of AI innovation foreshadows an evolution in business operations and decision-making. Instead of solely focusing on what happened or what might happen, AI will illuminate the actions that lead to better business outcomes.
This is the next step in the AI evolution, one that will empower your business to shift from human-initiated, tech-assisted processes, to AI-initiated, human-supervised actions. In this world, AI is the new BI.
That’s why today, we are thrilled to officially announce the public availability of Spotter, ThoughtSpot’s AI Analyst. This launch marks our first major milestone towards delivering on the promise of an agentic interface for an autonomous business.
Meet Spotter, your new AI Analyst
While business questions might look simple on the surface, real-world data is often messy, complex, and full of nuance, making the journey from data to actionable insights difficult to navigate and hard to decipher. With Spotter as your AI Analyst, anyone can discover decision-ready insights on any dataset, for any question—all while ensuring enterprise-grade governance and trust. Let’s take a closer look.
Ask, answer, and take action
Spotter employs a familiar conversational interface. Now, asking questions is just as easy as messaging an analyst colleague. Aided by Spotter’s automatic quickstart suggestions, you can simply start by asking Spotter questions directly from the homepage. You can also ask questions on existing Liveboard visuals and KPIs, and Spotter brings the context of those analyses straight into your conversations.
During your conversation, you can find even more granular answers in three ways:
Ask a follow-up question using Spotter’s conversational AI
Toggle to token-based search for column-level precision
Drill anywhere on AI-generated answers to discover the finest level of granularity
When you’re done, easily save and share your answers to spread knowledge and positive business outcomes across your teams, departments, and organization.
Intelligent AI that learns with you
Spotter provides reliable, replicable, contextually rich insights for all users by learning and adapting to your business data. Training starts with data teams building ThoughtSpot Models, which establishes a domain-specific glossary for the LLM that reflects your unique business context. From there, training is as simple as engaging with Spotter—asking questions and applying insights. Spotter learns as we use it.
Because Spotter converts natural language questions into search tokens, answers are explained in terms that anyone can understand, correct, and modify. Spotter’s advanced human-in-the-loop (HITL) feedback system empowers every user to improve insight accuracy—no complex queries or code required. Data teams can then monitor user feedback and incorporate system learnings, bridging the gap between AI and business needs to ensure users get reliable and contextually relevant answers.
Embed AI at the point of impact
Spotter is part of ThoughtSpot’s complete platform, meaning it’s available via our web API and plug-and-play SDK. Now, you can easily embed conversational AI and agentic analytics experiences into your apps and custom agents, creating highly engaging user experiences and new revenue streams with a faster time-to-market.
And with accessible, actionable insights available at the point of impact, you can finally harness the true value of your data to build better outcomes for your business, customers, partners—or any other external stakeholders. Of course, Spotter is also available across mobile devices and embeddable in your most popular workplace application. When you need answers, your AI Analyst is there.
What makes ThoughtSpot’s AI Analyst different?
Spotter isn’t just another chart generator or chatbot adding bulk to your workflow. There are enough of those already in the market. Our adaptable, trainable, self-improving AI Analyst is designed to help all users make better decisions—delivering AI to everyone.
While that has a nice ring to it, providing accessible, reliable AI to everyone is a lofty goal. To deliver on this promise, we’ve strategically designed Spotter’s architecture to account for the challenges of real-world business data as well as the shortcomings of AI. Here’s how it works:
Business-augmented reasoning
Spotter’s business-augmented reasoning for questions (BARQ) is the core reason why the AI Analyst fetches the right answer the first time. By augmenting generative AI (GenAI) with ThoughtSpot’s translation and training processes, Spotter is able to overcome the ambiguity of natural language, expressibility challenge of analytical queries, and complexity of real-world business data.
Translation
Upon submission, Spotter translates your natural language question into a data question expressed in ThoughtSpot search tokens—a taxonomy derived out of the data by our relational search engine. This is important because LLMs are built for translation—not SQL generation. In fact, there are a number of cited errors with text-to-SQL, all of which Spotter stopgaps.
Training
Similar to how your go-to analyst learns more about your unique data interests through each interaction, we’ve designed your AI Analyst to train on your questions, follow-up questions, and interactions with our advanced HITL feedback system. Spotter then uses that training to supplement standard GenAI reasoning techniques, layering on the parameters known about each unique user to deliver more relevant results.
AI trust
Businesses cannot bet on AI if they cannot explain what AI did. Spotter’s unique translation process serves dual purposes, solving yet another challenge of bringing AI to everyone: Trust.
We’ll start with the question itself. Using ThoughtSpot’s relational search tokens, Spotter verifies the question against the semantic model, making sure that the question is safe, secure, and conforms to your administrators’ established constraints.
By making these tokens available within the conversational interface, Spotter provides a user-friendly interface for any user to verify the query and result. That means each person can use natural language to verify their own results and submit training feedback. This addresses a major gap with text-to-SQL systems, where only a person proficient with data and SQL can verify if AI answered their question correctly—a process inaccessible for a typical business user.
Even better, each translation from natural language to relational search token is mapped and stored on file for analysts or account administrators to review and verify. Spotter eliminates AI’s black box, empowering every user to have trust in their data and take an active role in ensuring the accuracy of their insights.
Augmented analytics engine
Spotter’s augmented analytics engine is built using ThoughtSpot’s relational search—the tried-and-trusted technology behind ThoughtSpot search and SpotIQ. Here’s why it’s trusted: If relational search tokens are correct, SQL is 100% accurate.
This is where other LLMs and chatbots fall short. Because SQL is paramount to the accuracy of an AI-generated answer, we’ve taken great care in eliminating any room for error. Spotter’s accuracy and precision works by taking the question, layering on tokens, resolving the join paths, resolving the question—the ratios, growth, percentage, cohorts, etc—and then delivering the generated SQL to your data source to retrieve the accurate answer. All of this is clearly mapped, so you can always verify and trust the accuracy of your insights.
Dynamic action layer
When the answer comes back, it’s time for action! That’s where our dynamic action layer comes into play. Every chart, graphic, or visual presented by Spotter is interactive. Whether you click in, drill down, or follow up, there is no dead end. That means there is no limit to your curiosity. And that experience remains the same whether you view it on mobile, connect it with your own agents, or embed it in your own application.
Here’s the best part: ThoughtSpot is not tied to any ecosystem or LLM provider. Spotter is designed to work across all your business data, cloud data systems, on-prem data systems, apps, and files—an AI Analyst designed to work the way you do. We're committed to supporting your choice of AI for Spotter.
Building AI for the rest of us
As we mentioned, today’s launch of Spotter is only the first of many major milestones toward delivering on the promise of an agentic interface for an autonomous business. Today, you have the ability to answer the most complex questions you can imagine on your massive, real-world datasets.
"At the highest level, everybody is now able to leverage GenAI to make better decisions and drive meaningful outcomes. With Spotter, we’re delivering on our mission to make the world a more fact-driven place—for everyone. Spotter is AI for all of us.”
Soon, Spotter will have the ability to tell you about your dataset. It will have the ability to answer “Why” questions. And beyond that, eventually, it will even be able to take action on your behalf as well—this is what we mean by autonomous.
Connect with your personal AI Analyst today
We hope you’re as excited about the AI evolution as we are. Today, I’m inviting you to join us. You can take a guided product tour, or jump right in and try Spotter for yourself.