AgentSpot a means for ThoughtSpot users to build and share AI
With business decisions often a collaborative process but AI tools frequently used in isolation, the vendor's new workspace provides an environment for multiplayer AI.
The proliferation of chatbots and agentic AI assistants has boosted the productivity of individual employees. Enterprise work, however, often takes place across teams and systems that need to work together.
ThoughtSpot aims to break down the barriers that isolate AI agents within the enterprise.
On Thursday, the vendor introduced AgentSpot, a workspace that not enables ThoughtSpot customers to build agents using natural language prompts and allows them to share what they build so the agents can become assets for the whole enterprise.
AgentSpot provides an environment for what ThoughtSpot terms "multiplayer AI." The workspace enables business users to build agents on top of their organization's shared memory, business logic and governance, and share those agents so they foster collaborative decision-making.
Foundational capabilities include:
Company memory to give all agents the same understanding of the business.
Agent memory so creators can refine how an agent works with each adjustment reaching all users.
User memory so the agents learn individual's preferences and responses get more relevant to that user.
Agent-user memory to retain each user's history with an agent so each interaction doesn't begin from scratch.
"AgentSpot is a big one," Mike Leone, an analyst at Moor Insights & Strategy, told TechTarget. "ThoughtSpot users have mostly come to it for answers, and now agents can take those answers and do something with them in systems like their [customer relationship management] or Slack [channels]. The agents get their numbers through ThoughtSpot, so they match what people see in their dashboards."
Enabling users to share agents is not unique to ThoughtSpot, but ThoughtSpot is distinguished by ensuring each agent is built on the same memory, business logic and governance, he continued.
"Plenty of vendors let people build and share agents now, from the big AI assistants to CRM and analytics platforms," Leone said. "The difference is what the agent knows."
Breaking constraints
While chatbots and agents make workers more efficient, they often do so in isolation. According to a McKinsey report published in August 2026, chatbots -- which one user to hold a conversation in a confined environment -- remain the most widely scaled AI tools.
Similarly, agents developed by self-service business users such as marketers and financial analysts, while beneficial to the individuals who built them, often never leave their purview. Meanwhile, other marketers and financial analysts in the same organization develop their own similar agents to do the same work.
AgentSpot is a big one. ThoughtSpot users have mostly come to it for answers, and now agents can take those answers and do something with them.
Mike LeoneAnalyst, Moor Insights & Strategy
ThoughtSpot's new workspace is designed to reduce the reliance on single-player AI such as chatbots and isolated agents so employees can collaborate, sharing not merely answers from agents but building on one another's work to create tools their entire organization can use.
"Vendors far and wide are making their play into agentic AI and significantly changing their value proposition -- ThoughtSpot certainly has," William McKnight, president of McKnight Consulting, told TechTarget. "By combining four distinct layers of memory with enterprise-grade governance … it expands isolated chatbot usage into a reusable, organization-wide AI workforce."
ThoughtSpot's governed semantic layer provides a measure of differentiation even though providing collaborative capabilities as part of agentic AI offerings is not unique, McKnight continued.
"ThoughtSpot's real technical edge is anchoring agents to its mature, governed semantic search layer without requiring bespoke RAG pipelines," he said.
Although a valuable addition, AgentSpot is merely the logical progression of ThoughtSpot's platform, according to Francois Lopitaux, the vendor's senior vice president of product management.
ThoughtSpot provides Builder, an environment where non-technical users can build agents using natural language prompts and guided by an AI-powered assistant. AgentSpot advances that by enabling colleagues to work together to build agents that serve more than a single user.
"The way I see it is it's the next step in leveraging the data that you have in your company to automate work," Lopitaux said. "For a decade, we have been an analytics and BI [provider], and it was about giving you insights so you can take actions. Here, it's about creating autonomous agents grounded on the same data can do more. In a way, we are disrupting ourselves, and if we don't do it, someone else will."
Pricing based on outcomes, not token use
Beyond providing a workspace for building agents that can benefit more than just one individual or team, ThoughtSpot aims to help customers control their spending on AI by charging for AgentSpot based on outcomes rather than token use.
Lopitaux noted that when enterprises build AI tools, they have little idea how many credits or tokens will be used once the tools are in production. The number of credits and tokens can depend on the model being used and the complexity of the agent. With outcome-based pricing, if a user creates an AI workflow and runs it five times per week, it will cost them five credits.
"It's a nightmare, and it's slowing down adoption," Lopitaux said. "That's why we wanted to come out with a pricing that everyone can understand, that is practical, and that is not linked to a model."
Leone likewise noted that outcome-based pricing, which is gaining momentum among vendors despite complications related to defining what constitutes an outcome, could benefit ThoughtSpot users.
"The big advantage [of outcome-based pricing] is you pay for each piece of work an agent does, not for seats or tokens, so bringing the whole company onto a shared agent doesn't raise the bill," he said.. "It's catching on because once an agent is doing the work, counting the people who log in stops tracking the value."
ThoughtSpot customers can start using AgentSpot for free, with no limits on users but a weekly allowance of 25 credits for agent conversations and workflow runs. Organizations requiring more usage can pay for a pricing plan that comes with 250 credits per week.
"I'm hearing more about outcome-based pricing," McKnight said. "ThoughtSpot is on the leading edge of this, and it will be worth watching what they're doing. … With outcome-based pricing, AgentSpot is allowing companies to deploy unlimited users and agents without rationing access, while keeping building and testing no cost."
The next step: Moving AgentSpot into ThoughtSpot Embedded
While AgentSpot's effectiveness can only be measured once enterprises have been using the workspace, it appears appropriately constructed to enable users to build and share agents given the development foundation ThoughtSpot already has in place, according to Leone. As long as the underlying data is properly prepared, users should be able to collaboratively develop multiplayer AI tools.
"[ThoughtSpot's] builder is a big reason why," he said. "A business user describes what they want, and it writes the instructions most folks couldn't write themselves. The work that's left sits underneath, because agents give the most reliable numbers when a company has already modeled its data in ThoughtSpot."
ThoughtSpot's next step is to make AgentSpot part of ThoughtSpot Embedded so that customers can build agents, AI workflows and applications into their own products rather than just use the workspace in their ThoughtSpot environments," Leone continued.
"Embedded is a big part of ThoughtSpot's business. Those customers can already put Spotter's analyst agents in their products, so AgentSpot is the natural next piece," he said.
McKnight also noted that ThoughtSpot has provided a solid foundation for AgentSpot by anchoring AI development to governed semantic layers and organizational memory. However, he added that to address the performance of agents once they're in production.
"ThoughtSpot must continue to work on enterprise-grade agent evaluation harnesses to prevent non-deterministic drift and add transactional write-back safeguards with automated rollback and idempotency guarantees for third-party systems," McKnight said.
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.
