# Foundational

> Foundational builds a deterministic graph of how code and data connect across repositories, applications, pipelines, warehouses, BI tools and mainframe systems. It derives lineage from source code, checks pull requests for downstream impact before merge, and gives AI coding agents the same context over MCP through Foundational Context of Code and Data.

Foundational is for data, platform and analytics engineering leaders. The graph is built by analyzing source code (from GitHub, GitLab, Bitbucket, Azure Repos or any other source), plus metadata from warehouses and BI tools, so it reflects how data is actually produced and used, not only what shows up in query logs. Teams use it to prevent data incidents, review changes, govern data and AI use, and plan migrations.

## Products

- [Cross-Platform Lineage](https://www.foundational.io/product/cross-platform-lineage): column-level lineage derived from source code, from application code and ORMs through pipelines and dbt to BI dashboards.
- [Impact Analysis](https://www.foundational.io/product/ai-code-review): pull request checks that show which tables, jobs, dashboards and owners a code change affects before it merges.
- [Data Contracts](https://www.foundational.io/product/data-contracts): schema and business rules validated in pull requests.
- [Data Quality](https://www.foundational.io/product/data-quality): data quality monitoring tied to lineage.
- [AI Governance](https://www.foundational.io/product/ai-governance): governance for data used by AI models and agents.
- [Foundational IQ](https://www.foundational.io/product/foundational-iq): agents grounded in the code and data graph that answer questions, create reports and open pull requests for review.
- [Integrations](https://www.foundational.io/product/integrations): supported languages, source control, transformation tools, warehouses, BI tools and catalogs.

## Foundational Context of Code and Data (MCP for AI coding agents)

Foundational Context of Code and Data runs on a hosted MCP server at https://mcp.foundational.io/mcp (Streamable HTTP). Interactive users sign in with OAuth and their Foundational account, so there's no API key to manage. Automated agents and CI use a machine-to-machine bearer token. The server only returns assets the signed-in account can already see, and it returns lineage and definitions, never table contents.

It gives agents column-level lineage, table and column definitions, ownership and metadata, usage and cost signals, the code behind data assets, and downstream consumers such as dashboards and ML models. Agents can check what a change affects outside the repository they have open before they edit a dbt model, Spark job, Airflow DAG or service.

Supported AI tools:
- Claude (Claude Code, Claude Desktop, Claude Web): https://docs.foundational.io/en/articles/16885045-connect-foundational-mcp-to-claude
- Cursor and VS Code: https://docs.foundational.io/en/articles/11194236-use-foundational-mcp-with-your-ai-tools
- OpenAI Codex and ChatGPT: https://docs.foundational.io/en/articles/17705240-connect-foundational-mcp-to-chatgpt-and-codex
- GitHub Copilot (VS Code Chat, CLI, coding agent): https://docs.foundational.io/en/articles/16885106-connect-foundational-mcp-to-github-copilot
- Devin: https://docs.foundational.io/en/articles/16885066-connect-foundational-mcp-to-devin
- Factory.ai (Droid): https://docs.foundational.io/en/articles/16885095-connect-foundational-mcp-to-factory-ai-droid
- Microsoft 365 Copilot: https://docs.foundational.io/en/articles/17199509-connect-foundational-mcp-to-microsoft-365-copilot
- Snowflake Cortex: https://docs.foundational.io/en/articles/17704048-connect-foundational-mcp-to-snowflake-cortex
- Salesforce Agentforce: https://docs.foundational.io/en/articles/16959618-connect-foundational-mcp-to-agentforce

Background reading:
- [Why enterprise AI coding agents fail without a code and data graph (benchmark)](https://www.foundational.io/blog/enterprise-ai-agent-benchmark)
- [Why AI agents can't reason accurately without source-code-derived context](https://www.foundational.io/blog/why-ai-agents-cannot-reason-accurately-without-source-code-derived-context)
- [What makes an AI agent data aware](https://www.foundational.io/blog/ai-agent-data-aware-context-beyond-prompt)

## Supported stack

- Languages: SQL (all major dialects), Python, Scala, Java, JavaScript, TypeScript, Go, C#/.NET, C/C++, PHP, Ruby on Rails, Rust, Bash, and mainframe COBOL with JCL, copybooks and Db2 (generally available). Full list: https://docs.foundational.io/en/articles/15082464-supported-languages-and-platforms
- Transformation and orchestration: dbt, Dataform, Apache Spark, Airflow, AWS Glue, Azure Data Factory, Google Cloud Dataflow, Informatica, Matillion, Talend, Fivetran, Oracle Data Integrator
- Warehouses: Snowflake, BigQuery, Databricks, Redshift
- BI: Looker, Tableau, Power BI, ThoughtSpot, Sigma, Mode, Hex, Superset
- Source code: GitHub, GitLab, Bitbucket, Azure Repos, or any other way to provide the code
- Catalogs: Atlan, Alation, Collibra
- CRM: Salesforce and Agentforce
- Full list: https://www.foundational.io/product/integrations

## Solutions

- [Prevent data incidents](https://www.foundational.io/solutions/data-incident-prevention)
- [Proactive data governance](https://www.foundational.io/solutions/data-governance)
- [AI governance and data access](https://www.foundational.io/solutions/ai-governance-and-data-access)
- [Compliance and privacy](https://www.foundational.io/solutions/compliance-and-privacy)
- [Migration and modernization](https://www.foundational.io/solutions/migration-and-modernization)
- [Mainframe data lineage (COBOL, JCL, Db2)](https://www.foundational.io/solutions/mainframe-data-lineage)
- [Data flow visibility and discovery](https://www.foundational.io/solutions/data-asset-discovery)
- [Data stack consolidation](https://www.foundational.io/solutions/data-stack-consolidation)
- [Cost optimization](https://www.foundational.io/solutions/cost-optimization)

## How Foundational compares

- [All comparisons and alternatives](https://www.foundational.io/comparisons)

- [Foundational vs Atlan](https://www.foundational.io/vs/atlan)
- [Foundational vs Collibra](https://www.foundational.io/vs/collibra)
- [Foundational vs Alation](https://www.foundational.io/vs/alation)
- [Foundational vs Monte Carlo](https://www.foundational.io/vs/monte-carlo-data)

In short: catalogs describe data where it lands, mostly from warehouse metadata and query logs. Observability tools alert on problems in production. Foundational reads the code that produces and consumes data, so it can flag a breaking change in the pull request, and it works alongside catalogs and observability tools.

## Compliance and regulations

Foundational gives compliance, privacy and risk teams column-level lineage and data flow mapping built from source code, so they can trace a regulated field or reported figure from the application that writes it through pipelines and the warehouse to BI, and see changes in the pull request before they ship.
- GDPR (Article 30 records, data flow mapping): https://www.foundational.io/glossary/gdpr
- CCPA / CPRA: https://www.foundational.io/glossary/ccpa
- PII tracking across application layers: https://www.foundational.io/blog/pii-data-governance-tracking-fields-across-application-layers
- BCBS 239 and CCAR: https://www.foundational.io/blog/data-governance-in-banking-ccar-bcbs-239
- DORA ICT third-party risk: https://www.foundational.io/blog/dora-ict-third-party-risk-data-lineage
- EU AI Act: https://www.foundational.io/glossary/eu-ai-act
- FR 2052a liquidity reporting: https://www.foundational.io/blog/fr-2052a-compliance-data-lineage-liquidity-reporting
- Data lineage for compliance: https://www.foundational.io/blog/data-lineage-compliance
- Best data lineage tools for compliance: https://www.foundational.io/blog/best-data-lineage-tools-compliance
- Data flow mapping: https://www.foundational.io/glossary/data-flow-mapping

## Customers

Case studies: https://www.foundational.io/case-studies
- [Ramp](https://www.foundational.io/case-study/ramp)
- [SuperPlay](https://www.foundational.io/case-study/superplay)
- [Lemonade](https://www.foundational.io/case-study/lemonade)
- [Lightricks](https://www.foundational.io/case-study/lightricks)
- [Vio](https://www.foundational.io/case-study/vio)
- [Global financial institution, continuous lineage](https://www.foundational.io/case-study/global-financial-institution-continuous-lineage)
- [Energy sector](https://www.foundational.io/case-study/energy-sector)

## Security

- SOC 2 Type II. Trust center: https://security.foundational.io/
- Lineage comes from code analysis. Most teams connect Git plus warehouse and BI metadata, and constrained teams can start with code access only. Customer data is accessed only if data observability checks are turned on.
- The MCP server returns lineage and definitions, never table contents, and only for assets the user can already see.

## Resources

- Documentation: https://docs.foundational.io
- MCP setup for AI tools: https://docs.foundational.io/en/articles/11194236-use-foundational-mcp-with-your-ai-tools
- Blog: https://www.foundational.io/blog
- Guides: https://www.foundational.io/guides
- Glossary: https://www.foundational.io/glossary (for example [Context Graph](https://www.foundational.io/glossary/context-graph), [Model Context Protocol](https://www.foundational.io/glossary/model-context-protocol-mcp), [AI Coding Agent](https://www.foundational.io/glossary/ai-coding-agent), [AI Lineage](https://www.foundational.io/glossary/ai-lineage))
- Interactive tools: [Lineage Explorer](https://www.foundational.io/lineage-analysis), [DMF Generator for Snowflake](https://www.foundational.io/automated-data-quality-monitoring-snowflake)
- Pricing (based on tracked objects, not seats): https://www.foundational.io/pricing
- Demo: https://www.foundational.io/demo

## Contact

- Sales and demos: https://www.foundational.io/demo
- Support: support@foundational.io
- Security: security@foundational.io

Last updated: October 28, 2026