Solid Integration Setup Guide
What is Solid?
Enterprises are racing to adopt AI, but without a trusted understanding of their data, AI breaks at scale. Solid fixes this by giving AI a single, always-up-to-date source of truth so it can deliver reliable answers, power workflows, and act with confidence.
Trusted by Enterprise Teams
"Solid was the missing piece in helping us operationalize AI on our enterprise data. It turns the business logic already embedded across our analytics environment into governed, tested semantic models, so our teams can scale AI use cases easily without manual effort."
— Tal Homsky, CDO, Bank Leumi (~$300B Total Assets)
"Solid has given us a strong foundation where definitions stay aligned as our data and business evolve, so AI can deliver answers and power workflows we can actually trust."
— Meenal Iyer, SVP of Data, SurveyMonkey
What Solid Does
Once connected to your data infrastructure, Solid automatically:
- Discovers and catalogs all your data assets across platforms
- Tracks data lineage to show where data comes from and how it flows
- Monitors usage patterns to understand how your data is being used
- Extracts metadata from BI tools, databases, and warehouses
- Provides a unified view of your entire data landscape
This creates a trusted foundation that makes AI reliable and enables your entire organization to make confident, data-driven decisions.
How Integration Works
Setting up Solid follows a simple five-step process:
1. Create Service Account
Create a dedicated service account in your data platform (never use personal accounts for security and continuity).
2. Grant Permissions
Provide read-only access to metadata and query history. Solid accesses data when testing queries to ensure accuracy, but does not store your actual business data—only information about the structure and usage of that data.
3. Generate Credentials
Create authentication credentials (API tokens, service account keys, or Personal Access Tokens) based on your platform.
4. Connect in Solid
Add your credentials in the Solid interface:
- Navigate to Settings → Integrations
- Select your platform
- Enter connection details
- Test the connection
- Choose which resources to monitor
5. Monitor & Maintain
Solid automatically syncs on a schedule. Initial setup takes minutes, with the first sync and system calibration typically completing within a few days depending on your environment size.
Supported Platforms
Solid integrates with the tools your business already uses:
Data Warehouses
| Platform | Setup Guide | Key Requirements |
|---|---|---|
| BigQuery | BigQuery Setup Guide | Service account with BigQuery Metadata Viewer role |
| Snowflake | Snowflake Setup Guide | Service user with ACCOUNT_USAGE access |
| Databricks | Databricks Setup Guide | Service principal with Unity Catalog access |
| Amazon Redshift | Redshift Setup Guide | Service user with schema-scoped SELECT access |
| Microsoft Fabric / Synapse | Fabric / Synapse Setup Guide | Login/user with db_datareader role |
| Teradata | Teradata Setup Guide | Service user with schema-scoped SELECT access |
Databases
| Platform | Setup Guide | Key Requirements |
|---|---|---|
| SQL Server | SQL Server Setup Guide | Database user with VIEW SERVER STATE permissions |
| Oracle | Oracle Setup Guide | Read-only user with SELECT ANY TABLE and SELECT ANY DICTIONARY |
| IBM Db2 | Db2 Setup Guide | Read-only user with schema-scoped SELECT access |
Business Intelligence Tools
| Platform | Setup Guide | Key Requirements |
|---|---|---|
| Tableau Cloud | Tableau Cloud Setup Guide | Explorer user with Personal Access Token |
| Tableau Server | Tableau Server Setup Guide | Viewer user with Personal Access Token |
| Looker | Looker Setup Guide | Service user with API credentials |
| Power BI | Power BI Setup Guide | Service Principal with required permissions |
| Qlik Sense | Qlik Sense Setup Guide | Service account with API key or client certificate |
Other Tools
| Platform | Setup Guide | Key Requirements |
|---|---|---|
| JIRA | JIRA Setup Guide | Not currently offered |
Agentic Platforms (MCP Consumers)
Solid's semantic layer is exposed to any MCP-compatible AI agent platform, including:
| Platform | Notes |
|---|---|
| Snowflake Cortex | Solid can export to Snowflake Semantic Views for Cortex Analyst; also supports benchmarking Solid vs. Cortex Analyst on the same model |
| Microsoft Copilot | MCP-compatible; the Solid MCP server works with Copilot Studio |
| ChatGPT / OpenAI agents | MCP-compatible |
| Claude / Anthropic agents | MCP-compatible |
| CrewAI | Multi-agent orchestration framework with native MCP tool support |
| Cursor | MCP-compatible IDE; developers and analysts can call Solid MCP tools directly from the editor |
| Custom in-house agents (via MCP) | Any agent built to the MCP protocol specification |
| Bedrock AgentCore | MCP-compatible |
| UI Path | MCP-compatible |
See Getting Started with the Solid MCP Server to connect any of these.
What Data Does Solid Collect?
Solid collects metadata and usage information to build your data catalog:
- Structure metadata: Database schemas, table definitions, column details, and relationships
- Usage patterns: Query history, access frequency, and performance metrics
- BI metadata: Dashboard definitions, report structures, and data source connections
- Lineage information: How data flows between systems and transforms along the way
Important: Solid accesses data when testing queries to ensure accuracy, but does not store your actual business data—only information about the structure and usage of that data.
Required Inputs
| Input | Description | How Used |
|---|---|---|
| Schema & Metadata | Table definitions, column names, data types, constraints, indexes | Seeds semantic model table/column selection; PII scanning; information schema for SQL validation |
| Query Logs | Historical SQL queries executed against the warehouse | Query qualification funnel → relationships, metrics, certified SQLs, benchmark questions |
Optional Inputs (Enrich Model Quality)
| Input | Description | How Used |
|---|---|---|
| Data Samples (non-PII) | Small samples of column values from non-PII columns | Categorical value mapping for filters (e.g., "US" vs "USA"); improving SQL filter accuracy |
| Wikis & Documentation | Internal wikis, onboarding docs, PDFs, slide decks, Confluence pages | Glossary enrichment; understanding business context and terminology |
| BI Reports & Metrics | Tableau workbooks, Power BI reports, Looker dashboards, dbt semantic layer | Metric extraction; relationship inference; understanding which columns matter |
| Data Catalog exports | Collibra, Atlan, or other catalog exports | Richer semantic model generation; pre-built definitions and governance context |
What Solid Does NOT Ingest
- Raw production data (only schema/metadata and non-PII samples)
- PII columns (detected and discarded at ingestion time)
- Data from sources outside the configured access scope — Solid's access is scoped by you
- Non-relational warehouse content (e.g., document tables, semi-structured PDF storage) as queryable SQL data — PDFs and similar artifacts are used only as documentation/glossary context
See Collect and Offline Process for how these inputs are turned into a documented, ranked, searchable catalog.
Solid's Role in Your Architecture
Solid is not a replacement for your existing tools. It's a semantic generation, management, and enablement layer that sits between raw data and every consumer of that data.
| Tool in Your Stack | Solid's Relationship | Value Solid Adds |
|---|---|---|
| Data Warehouses (Snowflake, Redshift, etc.) | Connects to; reads metadata and query history; the DB Agent validates generated SQL | No data movement — Solid generates SQL only; execution happens with your credentials at runtime |
| Data Catalogs (Collibra, Atlan, etc.) | Ingests documentation and definitions as context for model generation | Richer semantic model generation; consistent terminology from catalog to model |
| BI Tools (Power BI, Tableau, Looker) | Ingests BI reports and metrics as signal during model generation; can push models back as an export | BI metrics become the source of truth for semantic models; models exported back for BI consumption |
| AI Agents (Bedrock, Copilot, Custom GPTs) | Exposes the semantic layer via MCP tools at runtime | Agents get accurate SQL and semantic context without needing to know your schema |
| dbt / Semantic Layers | Can ingest dbt definitions; can export generated dbt YAML back to your dbt repos | Automates what previously required manual dbt semantic layer authoring, plus adds testing |
Getting Started
Ready to give AI a trusted foundation? Here's what you'll need:
- Administrative access to your data platforms
- 15-30 minutes for initial setup per platform
- Access to documentation for your specific platforms
Detailed setup guides for each platform are available in our documentation.
Common Questions
How long does setup take?
Initial configuration takes 15-30 minutes per platform. The first metadata sync and system calibration typically completes within a few days for most environments, with subsequent syncs being incremental and faster.
Will this impact system performance?
No. Solid uses read-only access and queries metadata outside of peak hours. There's no impact on your production workloads.
What if we use multiple platforms?
Solid is designed to connect all your platforms into one unified view. Set up each integration separately, and Solid will automatically map relationships across systems.
Can we control what Solid accesses?
Yes. During setup, you select exactly which databases, schemas, projects, or folders Solid monitors. You can adjust these selections anytime.
What happens to our data if we disconnect?
You maintain full control. You can pause syncing, modify access, or completely remove an integration at any time through the Solid interface.
Support
Need assistance during setup?
- Documentation: Complete guides for each platform
- Platform-specific guides: Step-by-step instructions with screenshots
- Solid Support: Contact us at [email protected] with your integration type, error messages, and steps taken
We're here to ensure your integration goes smoothly.
Ready to unify your data landscape? Contact your Solid representative or visit our documentation to begin your integration journey.
Updated about 2 hours ago