Skip to content

Introduction

Every team has databases. PostgreSQL, MongoDB, BigQuery, MySQL, ClickHouse, Redshift — often several at once. The tools to work with them haven’t changed in years:

  • SQL clients (pgAdmin, DataGrip, DBeaver) give you a query editor and a schema tree. That’s it. No intelligence, no memory, no assistance.
  • ChatGPT/Claude can write SQL, but can’t see your actual schema, can’t execute queries, and hallucinate column names.
  • BI tools (Metabase, Looker) are for dashboards, not exploration. Building a one-off query means leaving the tool.

You end up copying schemas into ChatGPT, pasting queries into your SQL client, fixing errors, going back to ChatGPT — a manual loop that shouldn’t exist.

Mako is a SQL client where the AI is the interface.

  1. Universal Database Client — One console that connects to PostgreSQL, MongoDB, BigQuery, MySQL, ClickHouse, Redshift, SQLite, and Cloudflare D1. Switch between databases without switching tools.
  2. AI Agent with Real Access — The agent inspects your actual schemas, writes queries, executes them against your real data, reads the results, and iterates. No copy-pasting. No hallucinated column names.
  3. Console-First Design — The agent doesn’t just answer in chat. It places working queries directly in your console editor, ready to run, save, or modify.
  4. Learns Your Database — A self-directive system lets the agent build persistent knowledge about your schema, naming conventions, and common patterns. It gets better the more you use it.
  5. Collaboration — Multi-tenant workspaces with shared consoles, saved queries, and team management.
┌─────────────┐ ┌──────────────┐
│ React App │ │ Hono API │
│ (Vite) │──│ (Node.js) │
└─────────────┘ └──────┬───────┘
┌──────┴───────┐
│ │
┌──────┴──────┐ ┌────┴─────┐
│ Query │ │ AI Agent │
│ Runner │ │ (Vercel │
│ (9 drivers) │ │ AI SDK) │
└─────────────┘ └──────────┘