The walkthrough on this page covers 6 steps: 1. Request a demo or start signup 2. Model your data in the semantic layer 3. Ask questions through Analytics Chat 4. Explore and iterate in Workbooks 5. Build dashboards with the Dashboard Agent 6. Connect AI tools via the MCP server.

What are the limitations of Cube?

No pricing figures or plan tiers appear on the homepage, so budget planning requires a sales conversation. The product assumes a modeled semantic layer and a real data warehouse; teams without that foundation face substantial setup before AI features pay off. Embedded use cases lean on engineering work for APIs and custom UI, so it is not a no-code tool for non-technical teams.

Cube

Name: Cube
Author: Cube

Agentic analytics built on a governed semantic layer

Advanced API

Visit Cube Back to AI tools

Own Cube? Verify ownership to get a dofollow link

FeaturedHand-picked by aiwizard

AI Angels400+ customizable AI girlfriends for NSFW chat, voice calls, and roleplay

Candy AIExperience the future of companionship with AI friends who chat, role-play, and connect with you on a human level.

ourdream.aiA vast AI roleplay platform with millions of characters for chat, images, video and voice

What is Cube?

Cube is an agentic analytics platform built on a semantic layer, pairing natural-language Analytics Chat, workbooks and dashboards with embedded analytics and MCP access for AI agents, so every answer stays governed and consistent across humans, products and autonomous tools.

Cube is an agentic analytics platform that puts a semantic layer at the center of both human-facing business intelligence and AI-driven analysis. Every metric and dimension is defined once in a shared model, and every surface, whether a chat window, a workbook, a dashboard or an external AI agent, queries that same definition. The result is consistency: the same question returns the same governed answer regardless of who or what asks it. For data teams, the platform covers the working stack in one place. Analytics Chat accepts plain English questions and builds queries against the semantic model rather than guessing at raw tables. Workbooks combine SQL with a visual editor and a Workbook Agent that drafts and iterates on queries alongside the analyst. Dashboards are assembled conversationally through a Dashboard Agent, with performance handled by pre-aggregations and caching. Cube also exposes an MCP server so tools like Claude, ChatGPT and Cursor, plus a Slack agent that replies in-thread, can query company data with the same governance applied. For product teams, Cube offers embedded analytics through several paths: an Analytics Chat API for a fully custom experience, iframes for chat and dashboards as the fastest drop-in route, Creator Mode that lets end customers build their own workbooks and dashboards inside the host app, and core data APIs for teams that want to build their own interface. Multi-tenancy is part of the architecture, and permissions flow from the model through to each customer. The tool suits analytics engineers, data platform teams and SaaS companies shipping analytics to their own users. It competes with semantic layers and BI suites from dbt and Looker, and the pitch is narrow but real: mature governance for organizations where AI agents now touch production data.

How do you use Cube?

1Request a demo or start signup
Begin on the homepage to request a demo with the Cube team, or follow the get started link to create a Cube Cloud account. A demo is the faster route if you need to scope embedded analytics for customers.
2Model your data in the semantic layer
Define metrics, dimensions and joins against your warehouse in the semantic model. Every downstream surface, including chat and agents, resolves queries against these definitions.
3Ask questions through Analytics Chat
Use Analytics Chat to query data in plain English. The chat builds queries against the semantic model, so responses stay accurate and match what other users and agents see.
4Explore and iterate in Workbooks
Open a workbook to combine SQL and the visual editor, and use the Workbook Agent to draft and refine queries against the same governed model before saving results.
5Build dashboards with the Dashboard Agent
Assemble visual reporting conversationally through the Dashboard Agent, then rely on pre-aggregations and caching for fast load times. Each chart resolves against your semantic definitions.
6Connect AI tools via the MCP server
Wire Cube to Claude, ChatGPT, Cursor or your own agents through the MCP server, and add the Slack agent if the team wants answers in-thread. Governance and permissions carry through.

Pros and cons

Pros

Semantic model defines each metric once, so chats, dashboards, embedded surfaces and external agents all return consistent answersAI
Multiple embedded paths (Analytics Chat API, iframes, Creator Mode, core data APIs) let product teams pick the level of control they needAI
MCP server connects Cube data to Claude, ChatGPT, Cursor and custom agents, plus a Slack agent that answers in-threadAI
Dashboard Agent and Workbook Agent add conversational building and query iteration on top of the same governed modelAI
Pre-aggregations and caching target sub-second dashboard performance, and multi-tenancy is built in with permissions flowing from the modelAI

Cons

No pricing figures or plan tiers appear on the homepage, so budget planning requires a sales conversationAI
The product assumes a modeled semantic layer and a real data warehouse; teams without that foundation face substantial setup before AI features pay offAI
Embedded use cases lean on engineering work for APIs and custom UI, so it is not a no-code tool for non-technical teamsAI
The AI surfaces are tied to Cube's definitions, and quality depends on how well the model is maintained over timeAI
Some integrations are described as on the way, so coverage beyond the documented MCP and Slack paths should be verified for a specific stackAI

How much does Cube cost?

Support

Cube runs a public Slack community with 12,830 members for questions, alongside documentation and a changelog. The homepage does not describe paid support terms.

Learn more

Integrations

An MCP server connects Cube to Claude, ChatGPT, Cursor and custom agents, and a Slack agent answers in-thread. The homepage states that more integrations are on the way.

Learn more

Features

Analytics Chat for natural-language questions grounded in the semantic model, Workbooks with SQL, a visual editor and the Workbook Agent, dashboards built via the Dashboard Agent with pre-aggregations and caching, an MCP server for Claude, ChatGPT, Cursor and custom agents, a Slack agent, and embedded options spanning the Analytics Chat API, iframes, Creator Mode and core data APIs.

Learn more

Frequently asked questions about Cube

Does Cube have an API?
Yes, Cube offers an API.
How do you use Cube?
The walkthrough on this page covers 6 steps: 1. Request a demo or start signup 2. Model your data in the semantic layer 3. Ask questions through Analytics Chat 4. Explore and iterate in Workbooks 5. Build dashboards with the Dashboard Agent 6. Connect AI tools via the MCP server.
What platforms does Cube support?
Cube is available on Web App.
What does Cube integrate with?
An MCP server connects Cube to Claude, ChatGPT, Cursor and custom agents, and a Slack agent answers in-thread. The homepage states that more integrations are on the way.
What are the limitations of Cube?
No pricing figures or plan tiers appear on the homepage, so budget planning requires a sales conversation. The product assumes a modeled semantic layer and a real data warehouse; teams without that foundation face substantial setup before AI features pay off. Embedded use cases lean on engineering work for APIs and custom UI, so it is not a no-code tool for non-technical teams.