Count review: AI analytics and research workspace that shows sources and assumptions for every claim

What Count does

Count is a workspace for analytics and research that combines an AI agent with BI-style reporting. Its main idea is that each fact it produces comes with an explanation of how it was made, listing sources, assumptions and inferences, so a reader can check the work before acting on it.

Teams can connect databases, files and business tools, including through MCP, and then build slide decks, reports, dashboards and interactive apps on a shared canvas. Colleagues can comment, tag each other and work through answers together in real time. Editors can work with SQL, Python, low-code and the AI agent. Count also offers a semantic layer for defining business metrics, and a Slack integration for asking the agent questions.

Pricing is per editor. The Free plan costs $0, needs no card, and allows up to 5 users, 1 project and 3 canvases, with daily AI credit limits and limited models. The Pro plan costs $49 per editor per month, with viewers free, unlimited AI credits on a fair-use basis, unlimited canvases and projects, 3 database or MCP connections and 4GB compute capacity. Enterprise has no published price ("Let's talk") and adds unlimited editors, a REST API, SSO, SCIM, HIPAA and EU/US data residency, plus up to 32GB compute capacity. The pricing page does not state a yearly-billed price or a free trial.

• Reasoning chains showing sources, assumptions and inferences
• Real-time collaboration on canvases
• Reports, dashboards, decks and interactive apps
• Semantic layer and metric definitions
• Database, file and MCP connections
• SQL and Python support
• Slack integration for the AI agent
• Shared agent skills and context

Our take

Count fits teams that need analysis or research they can audit, such as people who must show where each number or claim came from before a board or stakeholder meeting. The free viewer role makes it workable when many people read results but only a few build them. The trade-offs: editors are billed per seat at $49 per month on Pro, Pro has only 3 database or MCP connections, and features such as the REST API, SSO and row-level permissions sit on Enterprise, whose price is not published. The Free plan suits evaluation but is capped at 3 canvases and daily AI credit limits. The pages describe no native mobile or desktop apps, so it appears to be used in the browser, plus Slack.