Skip to main content

HatiData Documentation

The data warehouse that gives every AI agent a brain. Sub-10ms SQL, persistent memory, semantic triggers -- all through a single Postgres connection.

Quickstart → | API Reference → | GitHub →

What is HatiData

HatiData is a data warehouse built for AI agents. It speaks the Postgres wire protocol, understands Snowflake-compatible SQL, and delivers sub-10ms query latency -- so your existing queries, dashboards, and dbt models work without changes. Beyond SQL, HatiData gives agents persistent memory, semantic triggers, chain-of-thought logging, and branch isolation out of the box.

Why HatiData

Legacy Cloud Warehouses were designed for human analysts running batch queries. AI agents need something different:

  • Sub-10ms latency -- Agents make hundreds of queries per reasoning chain. Seconds-per-query latency breaks the loop.
  • Your VPC, your data -- The data plane runs inside your infrastructure. No data leaves your network.
  • Agent-native billing -- Per-agent metering, per-query audit trails, and reasoning chain tracking out of the box.
  • Drop-in compatibility -- Write SQL the way you already do. HatiData transpiles it to DuckDB automatically.

Three-Layer Architecture

HatiData scales from a developer laptop to a multi-tenant enterprise deployment through three layers:

LayerPriceData LocationBest For
LocalFreeYour machineDevelopment, prototyping, offline work
Cloud$29/monthHatiData-managedSmall teams, startups, CI pipelines
EnterpriseCustomYour VPCRegulated industries, large-scale production

How It Works

Every SQL query flows through a secure pipeline that handles access control, SQL transpilation, execution, column masking, metering, and audit automatically. Security and compliance are built into every query path -- not bolted on after.

For the full pipeline details, see Core Concepts.

Key Differentiators

  • Local-first -- Start with hati init and query immediately. No cloud account needed.
  • Agent-native -- Agents identify themselves via startup parameters. Per-agent billing, scheduling, and audit happen automatically.
  • Snowflake SQL compatible -- The transpiler handles function mapping (NVL, IFF, DATEDIFF, etc.), type mapping (VARIANT, TIMESTAMP_NTZ, etc.), and construct rewriting (FLATTEN, OBJECT_CONSTRUCT, etc.).
  • Secure by default -- Access control, column masking, metering, and audit built into every query path.
  • Three-tier caching -- Layered caching from in-memory to disk to object storage for minimal latency.

Open Source

The CLI, SDKs, and integrations are open source under the Apache-2.0 license:

Next Steps

Stay in the loop

Product updates, engineering deep-dives, and agent-native insights. No spam.