SourceTable Superagents – Spreadsheets Meet Databases & APIs

TL;DR

TL;DR: SourceTable Superagents connects spreadsheets to external databases, REST APIs, and MCP servers so AI can analyze your data without writing code.

Source and Accuracy Notes

• Official site: sourcetable.com
• Launch HN: Show HN: Superagents
• MCP Protocol: modelcontextprotocol.io

What Is SourceTable?

SourceTable is a spreadsheet application built for the modern data environment. It looks and feels like Excel or Google Sheets, but with one key difference: your cells can pull live data from anywhere on the internet.

Superagents is the feature that makes this possible. Instead of manually exporting CSVs, writing Python scripts, or building ETL pipelines just to connect your spreadsheet to an external API, Superagents handles it for you. You point it at a database or endpoint, and the data flows into your sheet in real time.

The team behind SourceTable has a clear philosophy: spreadsheets are the original thinking tool, and they should be able to talk to the rest of the modern data stack without requiring an engineering degree.

How Superagents Work

Connecting a Data Source

From inside SourceTable, you specify what you want to connect to:

• Databases — PostgreSQL, MySQL, SQLite, and anything with a JDBC/ODBC driver
• REST APIs — any endpoint that returns JSON
• MCP servers — Model Context Protocol-compatible services, which includes a growing ecosystem of AI tools and data connectors

Once connected, you can browse the schema or endpoint structure directly in the spreadsheet UI and select which tables, views, or response fields you want to import.

AI-Powered Analysis

Once your data is in the spreadsheet, you can ask the built-in AI to:

• Generate reports and summaries from live database rows
• Build financial models that update automatically when the underlying API changes
• Create visualizations from data that previously required a dedicated BI tool
• Chain multiple agents together to handle complex multi-step workflows

The AI operates on top of your spreadsheet data the same way it would operate on a document — you describe what you want, and it builds it.

Orchestrating Multiple Agents

The name “Superagents” comes from how the system coordinates task-specific agents. A single workflow can call on multiple specialized AI tools — one might fetch CRM data, another enriches it with company info, a third generates a forecast — and all of them pass results back into the spreadsheet for the final output.

This is described as “conducting” the agentic web: you define the overall objective, and the system decides which agents to invoke in what order.

Setup Workflow

Step 1: Create a SourceTable Account

Visit sourcetable.com and sign up. The free tier includes a limited number of Superagent runs, so you can experiment before committing.

Step 2: Define a Superagent

Inside your SourceTable workspace, click New Superagent. You will be prompted to:

1. Name the agent
2. Select the connection type (Database, API, or MCP)
3. Authenticate with your data source (standard username/password or API key)
4. Browse available tables or endpoints
5. Select the fields you want to bring into the spreadsheet

Step 3: Connect to an MCP Server

If you are connecting to an MCP-compatible service, SourceTable will auto-discover available tools. You can then use natural language prompts to ask the AI to invoke specific tools and write results back into cells.

For example: “Pull the last 30 days of Stripe charges and calculate total revenue by customer” will invoke the Stripe MCP tool, fetch the data, and populate the sheet — no SQL required.

Step 4: Schedule and Automate

Superagents can be set to run on a schedule (hourly, daily, weekly) so your spreadsheet always reflects the latest state of your external data source. You can also trigger them manually from the Superagent panel.

Deeper Analysis

Why This Matters for Data Teams

The modern data stack is fragmented. Your sales data lives in Salesforce, your financial data is in Xero, your product analytics is in Mixpanel — and getting all of it into one view typically requires either a data engineer, a third-party integration tool, or a lot of manual CSV handling.

SourceTable Superagents treats the spreadsheet as the integration layer. Instead of building a pipeline to a data warehouse, you build it to a spreadsheet. Anyone who knows how to use Excel can now work with live data from any connected source.

For analysts and operators who are comfortable with spreadsheets but not with Python or SQL, this is a meaningful productivity upgrade.

The “Agentic Web” Vision

SourceTable’s founder, Eoin, draws a comparison to the semantic web dream of the early 2000s — a world where all data is linked and queryable. Superagents is their version of that vision: instead of a global semantic web, you get a personal data web where your spreadsheet is the interface and AI is the query language.

The MCP (Model Context Protocol) integration is significant here. MCP is an emerging standard for connecting AI agents to external tools and data sources. By supporting MCP natively, SourceTable positions itself to tap into the growing ecosystem of MCP-compatible tools without needing to build individual integrations for each one.

Comparison to Alternatives

| Feature | SourceTable Superagents | Zapier/Make | Traditional ETL | |---|---|---|---| | No-code data connection | Yes | Yes | No | | MCP support | Yes | No | No | | AI-native data analysis | Yes | No | No | | Agent orchestration | Yes | Limited | No | | Spreadsheet-as-UI | Yes | No | No |

Zapier and Make are great for workflow automation, but they are not designed for data analysis inside a spreadsheet. Traditional ETL tools require engineering resources and are not suited for the ad-hoc, exploratory workflow that spreadsheets enable.

Practical Evaluation Checklist

• Does your data source expose a database connection, REST API, or MCP server?
• Do your end users prefer working in spreadsheets over dedicated BI tools?
• Do you need AI to analyze and model data from multiple sources in one view?
• Is the data refreshed on a schedule, or does it need to be near-real-time?
• Are your non-technical team members comfortable requesting data changes via natural language prompts?

If you answered yes to three or more of these, SourceTable Superagents is worth evaluating.

Security Notes

• Data source credentials are stored encrypted and scoped to your workspace
• Superagent runs can be restricted to read-only access by configuring the data source with a read-only user
• API keys and OAuth tokens are not exposed in the spreadsheet UI — they are referenced by name, not by value
• Audit logs track which agents accessed which data sources and when

FAQ

Q: What databases does SourceTable support?

A: SourceTable supports PostgreSQL, MySQL, SQLite, and any database with a JDBC or ODBC driver. For specific integrations, check the documentation at sourcetable.com.

Q: Can I use Superagents with a private database behind a firewall?

A: Yes — SourceTable supports SSH tunnel connections for databases that are not publicly accessible. You provide the SSH credentials when configuring the connection.

Q: How does pricing work for Superagents?

A: Superagent usage is billed per run. The free plan includes a monthly allocation of runs; higher tiers offer more runs and additional features like scheduling and team collaboration. Check the pricing page for current rates.

Q: What happens if my API rate-limits during a Superagent run?

A: SourceTable will retry with exponential backoff. If the source remains unavailable, the Superagent run is marked as failed and you can re-run it manually once the rate limit window has passed.

Q: Does SourceTable Superagents support write-back to data sources?

A: Currently Superagents are read-only by default. You can configure a Superagent to write results back to a connected database, but this requires explicit opt-in and uses a write-enabled connection profile.

Conclusion

SourceTable Superagents solves a real problem: the gap between where data lives (databases, APIs, MCP servers) and where people actually work with it (spreadsheets). By treating the spreadsheet as a universal integration layer and AI as the query engine, it makes live data accessible to the people who need it most — without requiring engineering resources.

If your team spends significant time exporting data, reformatting CSVs, or building one-off scripts to get external data into a spreadsheet, Superagents could eliminate most of that work. The MCP support also future-proofs it — as more AI tools adopt the MCP standard, the list of compatible data sources grows automatically.

You can try Superagents for free at sourcetable.com.