AgentSpot

• Building a Monday workflow that pulls pipeline data from a ThoughtSpot model and Salesforce and posts a pipeline digest to a sales Slack channel
• Watching a metric such as cash position and alerting finance when it drops below a threshold
• Running a scheduled new-hire onboarding workflow that provisions accounts, assigns checklists, introduces a buddy and schedules a 30-day check-in for each start date
• Briefing an account manager on a customer before a call from CRM, email, Slack and meeting notes
• Generating a quarterly regional performance app from a governed revenue model that each regional lead opens and sees only their own rows
• Putting a policy or IT help agent in a public Slack channel so it answers recurring questions where employees already ask them

AgentSpot's pitch is less about smarter agents than about getting agents out of one person's chat window: they are built by the people closest to the work, shared company-wide, and grounded in memory and governed metrics. For companies that already model their data in ThoughtSpot, that semantic layer is a real differentiator, because a revenue question resolves to the same definition and the same row-level permissions whether a human or an agent asks it. The DO credit model is also unusually easy to forecast, since you are not billed per seat or token and failed runs are free (a monthly reasonable-use token cap still applies), though at $1,650 a month for 250 credits a week the jump from the free plan is steep. The documented gaps are typical of a launch-week product but matter for governance buyers: per ThoughtSpot's docs, tool-call-only audit logs, no masking before data reaches the model, no per-user credit limits, no model choice and SaaS-only hosting. Start on the free plan with a scheduled workflow you would otherwise do by hand; if your data does not live in ThoughtSpot, compare it with assistants and agent builders closer to where your work already happens.

About AgentSpot

ThoughtSpot launched AgentSpot on October 22, 2026 as what it calls Multiplayer AI: one workspace where anyone in a company builds AI tools from a plain-language description and shares them with colleagues instead of each person rebuilding the same thing in a private chatbot. It has three building blocks. Agents are conversational: you chat with them, and they can query data, use connected tools and take actions. Workflows are built through conversation but run on their own on an hourly, daily, weekly or monthly schedule; the builder splits each one into specialist sub-agents, one per tool or data source, that run in parallel and end in an output such as a Slack post or an email, with a test run before it goes live and a run history afterwards. Apps are interactive dashboards and briefings generated from a prompt, combining live charts, KPIs and tables with LLM-written summary blocks, where each viewer sees only the data their permissions allow. The distinctive part is the data layer. Agents can work over ThoughtSpot data models through Spotter, ThoughtSpot's analytics agent, so questions resolve through the company's governed semantic layer, its metric definitions and row-level security rather than ad hoc text-to-SQL. Memory sits at four levels (company, agent, user, and each user's history with a given agent), and admins seed org memory with key facts and rules. Connectors are MCP-based: a reference list of more than 1,000 tools, platform-native Google Workspace and Microsoft 365 connectors, any MCP server an admin adds, and per-tool permissions so an admin can, for example, switch off Jira ticket creation across every agent. Admins can also connect an agent to a Slack channel, where it answers @mentions or, if enabled, replies on its own. Pricing is the other departure: you pay in Data to Outcome (DO) credits, one per new agent conversation, workflow run or app load, with building, testing and failed runs free and unlimited users on every plan. A free plan with 25 credits a week needs no credit card; Fleet costs $1,650 a month billed annually for 250 credits a week. The tradeoffs are those of a new product: agents cannot be scheduled or called inside workflows, workflows have no event triggers yet, ThoughtSpot's docs say the audit log records tool calls but not prompts or outputs and that data reaches the LLM without a masking layer, SSO and audit logs are Fleet-only, and it runs only as ThoughtSpot-hosted SaaS.