What is Count?

Count is a web-based collaborative analytics workspace built around a shared canvas where SQL, Python and an optional AI agent work on connected data. Its agent builds charts, reports, decks and dashboards, and can also be reached from AI clients through MCP.

Who it's for

Count's homepage points to analysts, data teams, researchers and finance teams doing exploratory analysis, research and stakeholder reporting where mistakes are costly. Viewers are free, so it can suit teams where many people read results and few build them.

Pros & cons

Strengths

Count's homepage says the agent flags uncited claims, assumptions and model inferences, and lets teammates comment on results.
Count lists connectors for BigQuery, Snowflake, PostgreSQL and Databricks, tools such as Slack, Notion, HubSpot and Salesforce, and file uploads (PDF, DOCX, XLSX, PPTX, CSV, JPG).
Count says AI clients such as Claude, Claude Code, ChatGPT, Cursor and Gemini can connect through MCP, and lists optional GitHub sync.
Count's docs say the agent can access only database metadata (column names, types, descriptions) and does not query your database without explicit user approval.
Count's docs and Terms of Use say neither Count nor its AI partners train models on your data, and that AI input and output are your data.
Count's pricing page lists SSO, unlimited version history and REST APIs for Enterprise, and its Trust Center says paid customers can request a SOC 2 audit report.

Limitations

Pricing is per editor: Owners, Admins and Editors are billable seats. Viewers and Report Viewers are free. The FAQ says Admins are available on Enterprise.
The Free plan is capped at 1 project, 3 canvases, 5 users, 1 database or MCP connection and 2 automations, with a daily AI credit limit, limited models and 500MB compute. It has no scheduled reports or version history.
Pro is capped at 75 users and 3 database or MCP connections, with 14 days of version history. SSO and REST APIs are listed for Enterprise only.
The pricing page disagrees with itself: the Pro list says 4GB compute while the comparison table says 2GB, Pro AI credits are called unlimited in one place and fair use in the table and FAQ, and HIPAA is called certified in the FAQ but listed only for Enterprise in the governance table.
Count's Terms of Use say subscriptions are fixed-term, paid in advance and auto-renew unless you give notice, and that termination does not entitle you to a refund of prepaid fees. Count's liability for Count AI is capped at $100 USD, and the terms say AI output is not verified by Count.

Plans and pricing

Free: $0. No card required. 1 project, 3 canvases, up to 5 users, 1 database or MCP connection, 2 automations, daily AI credit limit with limited models, 500MB compute, web search and file uploads, community support.
Pro: $49 per editor a month. Viewers free. Up to 75 users, unlimited projects and canvases, 3 database or MCP connections, unlimited scheduled reports and automations, 14-day version history, AI credits on fair use with optional top-ups, email, Slack or Teams support.
Enterprise: Custom. Contact sales. Unlimited editors, unlimited connections with admin approval, unlimited version history, SSO, REST APIs, compute up to 32GB, onboarding and custom training.

Prices were read from Count's official pricing page. It shows only a monthly per-editor price for Pro and does not mention annual billing, refunds or storage limits. Confirm on the live pricing page, especially compute limits, AI credit wording and Admin availability, where the page conflicts.

Details checked October 2026 against the vendor's own pages. Prices and limits change, so confirm before you buy. Official pricing page ↗

Sources

This profile was compiled from these public pages. We have not run hands-on tests.

Summary

Count — at a glance

Try Count →

We couldn't find that tool

Count

At a glance