An Airtable alternative for people who want the data on their own desk

Why people leave Airtable

Per-seat pricing climbed from free to $20-45 per user, per month. A solo operator pays $240-540 a year just to keep using their own data.

The free tier caps you at 1,000 records per base. The next tier raises the limit to 50,000, but by then you are paying.

API access is rate-limited to 5 requests per second. Automation runs are capped per month. Both limits tighten on lower tiers.

Your data lives on Airtable's servers. If the company changes terms, raises prices, or sunsets a feature, your options are export-and-start-over.

What Dosirak does instead

A Dosirak project is a folder. TOML defines the schema and views. SQLite stores the records.

There is no record limit. SQLite handles millions of rows on a single machine.

There is no server. The app reads the folder directly. Sorting, filtering, and search run at disk speed with zero network round-trips.

You can back up the folder, put it in Git, or copy it to another machine. The data format is open. TOML and SQLite are standards with decades of reader support.

Airtable and Dosirak, side by side

The agent angle

Airtable offers a built-in automation builder: triggers, conditions, and actions you wire together in the browser. It works well for webhook-to-webhook flows.

Dosirak takes a different approach. An external agent (Claude Desktop, Cursor, or any MCP client) reads the schema, queries records, and files updates through the embedded MCP server. The agent runs locally, with no API key management and no per-request rate limits.

The difference is the approval step. When the agent needs a decision, it files a todo card linked to the specific record. You see the card in the desktop app, review the proposed change, and approve or reply. Nothing lands without your sign-off.

More on this in the MCP database feature page.

Moving a base over

1. Open the base in Airtable.
2. Export each table to CSV from the grid view.
3. Create matching tables in Dosirak with corresponding field types.
4. Import the CSV records.

What carries over: text, number, date, single select, email, URL, phone, and currency fields. Checkbox maps to yes/no.

What requires manual work: linked records need to be recreated as relationships. Attachments need to be downloaded and re-attached. Formula fields become calc expressions. Airtable automations do not transfer. The agent workflow is a different model.

Where Airtable still wins

• Real-time multiplayer editing. Ten people in different cities editing the same base simultaneously in a browser. Dosirak does not do this.
• The browser interface. Airtable works from any device with a browser. Dosirak is a desktop application.
• The integration marketplace. Airtable connects to hundreds of third-party services (Slack, Stripe, Salesforce, Zapier) out of the box.
• Granular permissions. Per-table and per-view access controls for teams. Dosirak is a single-operator app.

Dosirak does not yet support collaborative cloud editing for distributed teams. It is for operators who want their data on their own desk, who do not want to pay per seat for access to their own records, and who work with an AI agent that benefits from structured local schema.

Questions