dbt Coalesce '23 - Presentation

Universal Semantic Layer

Embed analytics, connect BI tools, and serve AI agents

OUR MISSION

Power the next generation of data-driven applications by making cloud data accessible and consistent to every data consumer in the enterprise

Universal Semantic Layer

Embed Analytics

Connect BI Tools

Serve AI Agents

Many-to-Many = Data Chaos

Data consumers are demanding all the data apps

Data engineers are juggling all the sources and outputs

How Cube fits into the stack

Custom Uls / Frontend React

Bls / Dashboards Metabase, Superset

Semantic Layer Cube

Notebooks HEX, Jupyter

Data Apps Streamlit, Dash

15

15

Key parts of the semantic layer

Data Modeling — Code-First, Dataset-Oriented

• Easy-to-grasp, LookML-like, open-source data modeling language
• Based on YAML, Python, and Jinja — or JavaScript, if you prefer that
• Dataset-oriented data modeling with cubes and views

Views

Cubes

orders

line_items

B1 #1

customers

users

B1 #2

orders

workspaces

companies

Data Modeling — Cubes & Views

Cubes are used to model entities and relationships between them. Similar to views in LookML.

Views are used to organize entities into exposable datasets. Similar to explores in LookML.

Access Control — Robust, Infinitely Flexible

• Row-based, column-based, role-based, use case-based access control
• Flexible, full-fledged support for multi-tenant applications
• Access control configuration kept as part of the data model

Caching — Cost-Saving, Blazing-Fast

• Multi-level caching enabling aggregate-aware, fast querying (10..300 ms)
• Flexible, scheduled or externally orchestrated refresh
• Caching configuration kept as part of the data model

APIs — Familiar, Diverse, Visualization-Agnostic

• SQL API and Semantic Layer Sync for self-serve/internal BI use case
• REST and GraphQL APIs for embedded analytics and real-time analytics
• OpenAI and LangChain integrations for AI applications

APIs — SQL/Rest/GraphQL APIs & SDKs

Unlimited, visualization-agnostic connectivity options:

APIs — SQL/Rest/GraphQL APIs & SDKs

Unlimited, visualization-agnostic connectivity options:

APIs — SQL/Rest/GraphQL APIs & SDKs

Unlimited, visualization-agnostic connectivity options:

Cube Cloud: More than just hosted Cube OSS…

Consistent metrics with Cube

Consistently calculated The data model within the semantic layer is the source of truth for metrics’ definitions.

Consistently refreshed The data model controls how and when metrics are refreshed. Downstream consumers never get stale results.

How Cube fits into the stack

Custom Uls / Frontend React

Bls / Dashboards Metabase, Superset

Semantic Layer Cube

Notebooks HEX, Jupyter

Data Apps Streamlit, Dash

Embed Analytics

Connect BI Tools

Serve AI Agents

Embedded analytics in just 2 weeks

90% less downtime 5x faster

Talk to your (modeled!) data

Security Scorecard replaced Sisense for a much faster, more integrated, data experience for customers.

Cloud Academy fixed their in-app analytics for customers, then connected their Preset - delivering product usage , NPS, and engagement internally.

Delphi built a conversational slackbot for true self-serve. Leadership can easily ask revenue attainment, sales progress, and other mission-critical KPIs.

Experienced Leadership & Trusted by 1000s

16,000+stars on Github 8,500+Slack community members 4,000+ active projects 10 Million+ downloads

Artyom KeydunovCEO Original creatorof Cube

Pavel TiunovCTO Original creatorof Cube

Jen GrantCOO Box, Elastic, Looker

14

15

15

Talk to our Expert Team

Scan the QR Code. Talk to Sales—Get a Personalized Demo. Ready to upgrade your data stack? Cube’s fast-to-market semantic layer helps companies build consistent, secure, and performant data apps.

• Deliver truly native embedded analytics.
• Build fast, scalable, cost-effective apps.
• Connect many sources to many front-end tools