Atlas vs WrenAI

WrenAI is an open-source text-to-SQL tool with a UI-based semantic modeling layer. Both Atlas and WrenAI use a semantic layer to improve SQL accuracy, but they take different approaches to modeling, deployment, and extensibility.

Quick Comparison

Semantic Layer

Atlas uses a code-first YAML semantic layer. Entity files, metrics, glossary terms, and query patterns live in your repository alongside your application code. Changes go through pull requests, are reviewed by your team, and deploy with your app. Atlas also provides a web-based semantic editor with schema-aware autocomplete and version history for teams that prefer a UI. The dynamic learning layer (atlas learn CLI + runtime learned_patterns with admin review) enriches the semantic layer from actual query patterns.

# semantic/entities/orders.yml
table: orders
dimensions:
  total:
    type: number
    description: Order total in USD
    sample_values: [29.99, 149.00, 499.50]

WrenAI uses a visual modeling interface where you define tables, relationships, and calculated fields through a UI. The modeling language (MDL) captures these definitions. This is easier for non-developers to get started with but harder to version control or review in code.

Trade-off: Atlas now offers both code-first YAML (git-versioned) and web editor (UI-based with version history) — covering both the engineer and analyst personas. Atlas's learning approach produces auditable YAML diffs, while WrenAI's modeling is more tightly integrated with its DataFusion query engine. WrenAI has more database connectors (12+ vs 7).

Embedding

Atlas is designed to be embedded in other applications. You can add a chat widget to any page with a script tag, use the @useatlas/react component for tighter React integration, or call the @useatlas/sdk API directly with streaming support. Atlas also provides 8 chat platform integrations (Slack, Teams, Discord, Telegram, Google Chat, GitHub, Linear, WhatsApp) via Chat SDK, a notebook interface for exploratory analysis, and a plugin marketplace for extensibility. For teams that don't want to self-host, Atlas Cloud provides a managed SaaS with 3-region deployment.

WrenAI offers API access for embedding text-to-SQL capabilities into other applications, but does not provide a pre-built widget or React component for drop-in integration. You would need to build your own UI layer on top of WrenAI's API. WrenAI does not offer chat platform integrations or a plugin marketplace.

Licensing

Both Atlas and WrenAI use AGPL-3.0 for the server. The difference is that Atlas's client libraries (@useatlas/sdk, @useatlas/react, @useatlas/types) and all plugins are MIT — you can embed them in commercial products without copyleft concerns.

WrenAI is fully AGPL-3.0, which requires that modified versions of the software must also be released under AGPL-3.0, and that users interacting with it over a network must be offered access to the corresponding source code.

When to Choose WrenAI

• Your semantic layer is owned by analysts or business users, not engineers
• You want a standalone text-to-SQL product, not an embeddable component
• You prefer a visual modeling experience over YAML files
• AGPL-3.0 is compatible with your use case