Skip to main content
Credible is a distributed platform designed for modern, developer-first workflows that prioritize reliability, scalability, and security. To achieve this, our architecture is built on a clear separation of concerns, primarily between the Control Plane for administration and the Data Plane for querying. This design ensures that managing your data models and running live queries are completely independent processes. Administrative operations like publishing a new model version or updating permissions are isolated from the high-throughput query workloads, allowing each plane to scale and operate independently.

​
The Core Components

Our architecture is composed of three primary layers: the Application Layer, the Infrastructure Layer, and Global Services. This structure provides a durable foundation for everything from local development to enterprise-scale deployment.

​
1. The Application Layer

This is where you and your team interact with Credible’s user-facing tools. The Application Layer is a consumer of our infrastructure APIs and includes services for:
  • AI-Powered Workspaces: Agentic dashboard builders, natural language workbooks, and other AI services for analysis and query generation.
  • Indexing & Retrieval Services: Manages the context needed for AI-assisted modeling and exploration, exposed via the Retrieval API.
All application services are accessed through a dedicated hostname, like acme.app.credibledata.com.

​
2. The Infrastructure Layer

The infrastructure is the engine of the Credible platform, responsible for managing your semantic models and serving data consistently. It is divided into two distinct planes, each with a dedicated API.

​
Control Plane: Administration & Governance

The Control Plane handles all administrative functions. It’s where you manage the lifecycle of your data assets and the security of your organization.
  • What it does: The Admin API provides programmatic access to the entire resource hierarchy, allowing you to automate and integrate your data governance workflows.
  • How you access it: Through the Admin Web App or programmatically via the global administrative API endpoint: <org>.admin.credibledata.com.

​
Data Plane: Querying & Data Access

The Data Plane is the high-availability workhorse for all data requests. It has one job: to serve consistent, trusted answers from your published semantic models.
  • What it does: The Data API provides access to your compiled, versioned Malloy packages. It is optimized for low-latency query execution and high concurrency.
  • How you access it: All data access is handled through a single, globally load-balanced API endpoint for maximum availability: <org>.data.credibledata.com.

​
Intelligent Context Management

A core capability of the Credible platform is its ability to build a comprehensive understanding of your business to power the entire AI-assisted workflow - from model creation to analysis to evolution.

​
1. Indexing for AI-Assisted Modeling

The first index is designed to understand your raw data landscape before a semantic model exists. Its purpose is to provide AI assistants with the deep context needed to help you create a robust model from scratch by synthesizing a rich contextual data from sources like database metadata, SQL query logs, existing semantic models, and unstructured documentation.

​
2. Indexing for AI-Powered Analytics

Once you publish a semantic model (a Malloy Package), a second, more focused index is created. This index is optimized for consumption and analysis and contains the model’s sources, dimensions, measures, and predefined views or queries. It can be configured to also index dimensional values, allowing users to ask questions and find the right query based on the data itself.

​
The Semantic Hub: Governance, Efficiency, and Trust

Because every query—from a human analyst or an AI agent—is routed through the semantic layer, Credible becomes the central nervous system for your organization’s data. This provides a single point of leverage for visibility and control.

​
Centralized Governance and Compliance

By serving as the single gateway for data access, the platform allows you to enforce fine-grained access controls (ACLs) consistently. Every query is logged, creating an immutable audit trail for security and compliance tracking.

​
Aggregation, Transparency, and Efficiency

The platform gains a unique, holistic view of all data consumption, allowing it to act as an intelligent aggregation layer. This means you can:
  • Optimize Performance and Cost: The platform tracks query cost and latency, using this data guide caching, pre-aggregation, and materialization strategies. This ensures queries are fast and cost-effective.

​
A Virtuous Cycle of Improvement

This centralized view enables a powerful feedback loop. By analyzing query history against the model, we can:
  • Identify Data or Model Coverage Gaps: Pinpoint common business questions that are not easily answered by the current model and suggest improvements.
  • Detect inconsistent usage patterns: Identify when different teams query the same concepts differently, revealing semantic drift or misaligned business definitions that should be standardized in the model.
  • Surface Data Silos: See where data is underutilized or disconnected from the core business model.
This constant cycle of analysis and feedback ensures your semantic model evolves with your business, becoming more durable, efficient, and trusted over time.
⌘I