Secure AI Collaboration Platform
Enterprise-grade infrastructure with agentic context creation for AI-powered data analysis
Key Features
Secure Kubernetes Environments
Every workspace runs in an isolated Kubernetes environment, providing enterprise-grade security and resource management. Your data and operations are completely contained within your own secure namespace.
- Isolated Namespaces: Each workspace runs in its own K8S namespace with strict resource quotas and network policies
- Enterprise RBAC: Fine-grained role-based access control for teams and individuals
- Audit Logging: Complete audit trail of all operations and data access
- Compliance Ready: SOC 2, GDPR, and HIPAA compliant infrastructure
Command Line Tools & Data Operations
Work with your data using familiar command-line tools and powerful in-memory data operations. DuckDB integration provides blazing-fast analytics without moving data.
- Full Shell Access: Use standard Unix tools like grep, awk, sed, and jq for data manipulation
- DuckDB Integration: In-memory SQL analytics with lightning-fast query performance
- Context Offloading: Automatically persist intermediate results to avoid context window pollution
- Script Execution: Run Python, Node.js, and other runtime environments securely
Federated Query Infrastructure
Connect to all your data sources through a single, unified interface. Our federated query engine enables seamless access across your databases and SaaS applications.
- Universal Connectors: Native integrations with PostgreSQL, MySQL, MongoDB, Snowflake, BigQuery, and 100+ more sources
- Cross-Source Queries: Join and analyze data across multiple databases in a single query
- Real-Time Sync: Live connections to your data sources with no ETL required
- Unified Interface: One query language to access all your data, regardless of source
Agentic Context Creation
Marco Polo automatically scans your data sources to discover relationships, entities, and business concepts. The AI receives rich, structured context about your data without manual documentation.
- Automatic Schema Discovery: Scan databases to identify tables, relationships, and foreign keys
- Entity Recognition: Detect business entities like customers, products, orders, and their connections
- Relationship Mapping: Build a graph of how your data entities relate to each other
- Semantic Understanding: Infer meaning from column names, data patterns, and usage to provide business context
Intelligent Context Management
Our smart context engine ensures the LLM always has the right information at the right time, without overwhelming the context window with irrelevant data.
- On-Demand Loading: Infrastructure details and schemas are fetched only when relevant to your query
- Selective Injection: Only the most pertinent business concepts and metrics are added to context
- Context Window Optimization: Efficiently pack relevant information while avoiding pollution
- Query-Aware Contextualization: Previous queries and results inform what context is needed next
Getting Started
Follow these steps to set up your secure workspace and start collaborating with AI
Add Marco Polo to Your AI Client
From Claude Desktop or any remote MCP-capable AI app, add the Marco Polo server as a custom connector.
Enter the following:
- Name: Marco Polo
- URL: https://mcp.marcopolo.dev
See the home page for installation instructions for other clients like VS Code, Cursor, and ChatGPT.
Connect and Authenticate
Click Connect, and you'll be prompted to authenticate with Google or GitHub. SAML and other authentication options are supported for enterprise customers.
Authentication Options:
- GitHub OAuth: Sign in with your GitHub account
- Google OAuth: Use your Google Workspace or personal account
- Enterprise SSO: SAML and other options available for enterprise customers
Your authentication is token-based and follows OAuth 2.0 best practices. We never see or store your passwords.
Marco Polo is Ready to Use
Once connected, you'll see Marco Polo in your list of MCP servers. You can now open a chat and start using it.
Connect Your Data Sources
The first time you use tools, you may need to approve them. When connecting to a data source, just follow the URL to enter credentials securely.
Supported Data Sources:
- Databases: PostgreSQL, MySQL, MongoDB, SQL Server, Oracle, and more
- Cloud Warehouses: Snowflake, BigQuery, Redshift, Databricks, Athena
- SaaS Applications: Salesforce, HubSpot, JIRA, Stripe, Shopify, and 100+ more
All credentials are encrypted at rest with AES-256 and in transit with TLS 1.3. Your data never leaves your control.
Start Querying Your Data
Once connected to data sources, your LLM can query them for you within our secure K8S infrastructure. Context around entities, schemas, and infrastructure is automatically created on-demand—with long-term memory persisting where applicable.
In this example, the LLM uses its command-line environment to write a SQL model for the query, executes it, and displays a preview of results. Tool calls can be expanded to see the exact SQL and commands used.
What You Can Do:
- Query any connected data source with natural language
- Join data across multiple databases using DuckDB
- Execute Python scripts for custom analysis
- Generate visualizations and reports
- Sync code and reports to Google Drive or GitHub
Your secure Kubernetes workspace includes DuckDB, Python, Node.js, and standard Unix tools—all isolated in your own namespace.
Available Tools & Capabilities
Your AI assistant has access to 9 tools organized by category through the MCP protocol
Shell Execution (3 tools)
🐚 Execute Command
Execute shell commands (git, ls, cat, head, grep, find, etc.) with configurable timeout. Perfect for text processing and data transformation.
📦 List Installed Packages
List available CLI tools and their versions in your workspace environment.
❓ Get Command Help
Get help documentation for any command to understand usage and options.
Data Sources (4 tools)
📋 List Data Sources
List all connected data sources (SQL databases, APIs, storage) with their capabilities and connection status.
📊 Query
Execute SQL queries or API requests against any data source. Results are automatically loaded into DuckDB for further analysis.
🗂️ Browse
List files in storage data sources (S3, GCS, etc.) with optional detailed metadata. Loads results into DuckDB.
⬇️ Download
Download files from storage data sources to local workspace scratch for processing.
Workspace Sync (2 tools)
⚙️ Configure Workspace Sync
Set up sync providers (GitHub, Google Drive) by connecting, disconnecting, or listing configurations.
🔄 Sync Workspace
Sync workspace files to/from external storage providers. Push code and reports to GitHub or save to Google Drive.
Connectors (1 tool)
🔗 Generate Connector URL
Generate OAuth or database connector setup URLs for secure credential configuration.
Enterprise Security & Compliance
🔐 Data Encryption
- AES-256 encryption at rest
- TLS 1.3 for data in transit
- Encrypted credential storage
- Hardware security modules (HSM)
👥 Access Control
- Role-based access control (RBAC)
- Multi-factor authentication (MFA)
- Single sign-on (SSO) integration
- Audit logs for all actions
🏢 Compliance
- SOC 2 Type II certified
- GDPR compliant
- HIPAA compliant
- ISO 27001 certified
🛡️ Infrastructure
- Kubernetes network policies
- Resource quotas and limits
- Automated security scanning
- Regular penetration testing