Query Streams

What it is

Query Streams is a secure SQL access layer. The homepage describes "secure data access for you, your team, vendors, apps, and AI": one outbound-only agent sits inside your network, and through it your data reaches Excel and Google Sheets, an AI analyst called Nova, MCP clients such as Claude and Cursor, your own apps over a REST API, and scheduled syncs, automations and alerts in Slack. It is for teams that live in spreadsheets and no-code tools but whose real data sits in SQL databases.

What you build and how

You do not drag blocks together here. You write SQL once, or let Nova write it, and then choose where the result goes. The homepage lists 3 steps. Deploy the lightweight Network Agent on a machine inside your network, with installers for Windows, Linux and macOS. Connect your databases: adding a connector auto-discovers schemas and tables for the query builder, and the page says 11 engines are supported out of the box, along with 8 API connectors. Then query from anywhere, whether from a spreadsheet, an AI assistant, an API call or a scheduled sync, without exposing SQL or credentials. The company's docs name the databases you add through the agent's local dashboard: Microsoft SQL Server, PostgreSQL, MySQL, MariaDB, SQLite, Microsoft Access, Oracle, BigQuery, Snowflake and DuckDB, with full SQL validation for each dialect. The navigation also lists a Query Builder, Share SQL Results and Dynamic SQL Filters.

That one design gives you several different outputs, which is why the product sits in several categories.

Live queries are the spreadsheet side. Pick a saved query in the Excel or Google Sheets add-in, hit Run, and refresh with one click, with no SQL to write or expose. The company's documentation says results stream to the worksheet on demand, with no data staging.

Automation is the data-sync side. The homepage says you can "push a query's result to Smartsheet, Airtable, Sheets and more on a schedule, every 5 minutes or any cron." An Italian-language guide on the site is more specific: it names Airtable, Smartsheet, Baserow, SeaTable, Google Sheets and Anvil.Works as the 6 platforms, and says each supports both append and full sync modes with cron scheduling. The guide frames this as building no-code database apps on real SQL data.

The Database REST API is the API-builder side. The homepage says you can go from database to a live REST endpoint in about 90 seconds and "turn any saved query into an API with a personal key per recipient," read-only and revocable, with no gateway to host. The site's tagline for this is one key per partner, no credentials shared.

Nova AI is an AI data analyst. You ask in plain English, and the site says Nova explores your schema, writes and runs read-only SQL, builds charts and graphs, and talks through the findings.

The MCP Server lets Claude, Cursor, ChatGPT and Grok query your database, "even one behind a corporate firewall." The company says no database credentials are ever stored in the MCP config.

Federated Queries run one SQL statement across several of your databases at once. The example is orders in MySQL joined to customers in PostgreSQL, where each database does its own share of the work and you get back a single table, with nothing exported or copied. The documentation also lists parallel query execution.

Alerts and Reports let you watch stock, margins or overdue invoices and send Nova-narrated reports to Slack, Discord or Google Chat.

The API Connector turns APIs into SQL tables. The homepage names Stripe, HubSpot and Shopify, and says the full dialect is available, unlimited, with no rate limits.

The homepage sums up the idea as writing SQL once, against one database or several at the same time, and then reaching it from wherever the team works, or letting it come to you the moment something changes through alerts. Nova is pitched as more than a question box: the page says it also helps you optimize queries on your own data. The homepage says the data is always governed, read-only and audited. Collaboration works through saved queries that colleagues can reuse without seeing the SQL.

What the site states

Pricing is partly public. The homepage offers a free forever tier with no credit card required. A comparison guide on the company's docs says the free tier has full feature access, including the on-premise agent, the Excel and Google Sheets add-ins, Nova AI and parallel query execution, with no row limits on any tier. It says paid plans start at $19 a month with flat pricing, no row limits and no per-database charges, and that you can connect multiple databases and run unlimited queries on any paid plan. The homepage also links a separate Team Pricing page.

On security, the stated model is the outbound-only agent: no inbound ports, no VPN, no firewall changes, and credentials that never leave. The Italian guide adds that data in transit is encrypted and destination API tokens are kept secure.

The homepage reports usage figures: 23,000+ queries a month, 850,000+ rows synced, 200+ organizations and 25+ countries. They are the company's own numbers. Support is listed in the navigation, along with docs and a contact page. The homepage embeds independent video walkthroughs by Luke Saunders covering live Excel and Sheets, building a REST API from your data and federated queries, and the site itself can be switched between English, Spanish, German, French, Portuguese, Japanese, Korean, Chinese, Italian, Dutch, Indonesian and Polish.

Who it suits and who it does not

Query Streams suits an operations or finance team that wants live database numbers in Excel or Sheets, or a team that runs its interface in Airtable or Smartsheet and needs production data underneath it. It also suits a company with databases behind a firewall that wants AI tools or partners to query them without handing out credentials, and a developer who wants a REST endpoint from a saved query in a minute or two.

It suits a founder who wants to build a customer-facing app less: it moves and exposes data, and the screens are the spreadsheets, apps and AI clients you already use. Because the access model is read-only, it is not a tool for writing changes back to a source database. Someone who needs transformations across many SaaS apps will find the connector list centered on databases. And the agent has to run on a machine you maintain, which is a small operational duty for the security it provides.