PRICING
Start Free. Scale to Enterprise.
Every tier includes the full ANDI substrate — persistent memory, CoT ledger, semantic triggers, branch isolation, and per-agent governance.
Local
Prototype on your laptop
$ hati init- 1,000 agent memories
- 100 semantic triggers
- 1 branch for isolation
- SQL + Vector search
- MCP-native (24 tools)
- 50 hybrid queries/day
- 384-dim embeddings
- Local query engine
- Open storage format
- Community support
Cloud
Ship agents to production
$ hati push --target cloud- 1M agent memories
- 1,000 semantic triggers
- 10 branches for isolation
- Managed MCP endpoint
- 10,000 hybrid queries/day
- 1,536-dim embeddings
- Dashboard + API keys
- Team collaboration
- Social login (Google, GitHub)
- Magic link sign-in
- Managed cloud infra
- Slack support
Growth
For agent fleets at scale
$ hati push --target cloud --tier growth- 10M agent memories
- Unlimited triggers
- 100 branches for isolation
- CoT Ledger (hash-chained)
- 100,000 hybrid queries/day
- 3,072-dim embeddings
- 3 environments
- Team RBAC (6 roles)
- Social login (Google, GitHub)
- Magic link sign-in
- Column masking
- Enhanced audit logs
- Shadow Mode report
Enterprise
For regulated agent deployments
$ hati push --target vpc- Unlimited memories
- Unlimited triggers
- Unlimited branches
- CoT Ledger (7yr retention)
- Bring Your Own Embeddings
- Unlimited hybrid queries
- In-VPC deployment
- Private Connectivity (AWS, GCP, Azure)
- CMEK encryption
- SAML / OIDC SSO
- Directory Sync (SCIM)
- Org-wide MFA Enforcement
- Row-level security
- Dedicated support SLA
- DPA/MSA/BAA templates
WHAT'S INCLUDED
Every Agent Feature. No Per-Operation Charges.
Memory, triggers, CoT ledger, and branches are included in every tier. You only pay for SQL compute.
CAPACITY CALCULATOR
How Big a Brain Do Your Agents Need?
Estimate your usage and see which tier fits.
COMPARE TIERS
Full Feature Comparison
FAQ
Frequently Asked Questions
Yes. No time limit, no credit card required. hati-local runs a local query engine with 1,000 memories, 100 triggers, and full MCP tooling. Free forever. When your agents need managed infra or team features, upgrade to Cloud ($29/mo).
Local is free for prototyping agents on your laptop. Cloud ($29/mo) adds managed infrastructure, 1M memories, and team collaboration. Growth ($199/mo) adds the CoT Ledger, 100 branches, RBAC, and fleet-scale capacity. Enterprise adds In-VPC deployment, PrivateLink, CMEK, and unlimited everything.
Yes. Run 'hati push --target cloud' when ready. Your data, schemas, and agent memories move seamlessly. No migration scripts.
Yes. Memories are namespaced per agent but can be shared across sessions, teams, and even agent types. An agent picks up exactly where it left off — or where another agent left off.
Agents get isolated schemas for speculative queries. They can explore, mutate, and test without touching production data. When done, merge the branch back or discard it. Zero-copy on create, copy-on-write on first mutation.
The CoT Ledger records every reasoning step with cryptographic hash chaining. Each entry is cryptographically linked to the previous one, making tampering detectable. With Enterprise 7-year retention and immutable storage, it is designed for regulatory and legal audit requirements.
HatiData is MCP-native and works with Claude, Cursor, Cline, Windsurf, LangChain, CrewAI, AutoGen, and the OpenAI Agents SDK. Any MCP-compatible client connects out of the box via 24 built-in tools.
HatiData combines SQL + vector search + chain-of-thought audit + semantic triggers in one system. No separate Pinecone or Chroma. No glue code between your database and your vector store. One local file, one query language.
With Local tier, on your machine. With Cloud, in our managed infrastructure. With Enterprise, in your cloud account (AWS, GCP, or Azure) — your VPC, your storage buckets, encrypted with your KMS keys.
Yes. pip install dbt-hatidata, update your profiles.yml, run dbt build. Our AI transpiler handles SQL dialect differences automatically.
Ready to give your agents a brain?
Start local in 30 seconds. Push to cloud when your agents need it.