# Omnia vs Letta

> Omnia’s memory redacts before storing, honours consent revocation, and sits in a platform that tests and releases the agent. In Letta, the agent itself decides what goes into its memory.

Category: Memory. Checked against Letta’s published material on 2026-10-22. Canonical page: https://www.altairalabs.ai/vs/letta/

## Letta’s agents manage their own memory. Omnia’s memory follows your privacy rules.

In Omnia, redaction, consent and erasure are set by the platform, and the agent can’t edit them. In Letta, from the MemGPT team, the agent decides what goes into its own memory.

## When it matters: An agent has talked with a customer for six weeks. The customer asks you to forget them.

**On Omnia:** Their card and phone numbers were redacted before anything reached long-term memory. Withdrawing consent covers what was stored over those six weeks, and the erasure runs as a data subject request with an audit trail. (Omnia docs: https://omnia.altairalabs.ai/how-to/privacy/configure-privacy-policies/)

**On Letta:** Background dreaming reviews recent conversations and rewrites memory, and its review step doesn’t ask for approval. Consent, PII redaction and erasure lifecycles aren’t documented. (source: https://docs.letta.com/configuration/memory/index.md)

## Memory with rules the agent can’t edit

Redaction and consent are set in the platform. Newer facts win by precedence and supersession, and consolidation runs on a schedule you set. *(enterprise)* (Omnia docs: https://omnia.altairalabs.ai/how-to/privacy/configure-privacy-policies/)

Letta: Agent-edited memory; in the legacy blocks, the last write wins. (source: https://docs.letta.com/guides/core-concepts/memory/memory-blocks/)

## Tested before it learns from customers

Before each release, simulated users talk to the agent through its real tools. A version that fails doesn’t go out. (Omnia docs: https://promptarena.altairalabs.ai/arena/explanation/scenario-design/)

Letta: Agent testing and versioned rollout aren’t documented. (source: https://docs.letta.com/agent-sdk/deployment/index.md)

## Runs on your Kubernetes

Omnia installs with Helm into your cluster, with workspaces isolating each team. (Omnia docs: https://omnia.altairalabs.ai/how-to/workspaces/manage-workspaces/)

Letta: Self-hosting uses a Dockerfile with Compose, Railway or Fly.io; Kubernetes isn’t documented. (source: https://docs.letta.com/self-hosting/index.md)

## At a glance

| Topic | Omnia | Omnia docs | Letta | Letta docs | Source |
|---|---|---|---|---|---|
| Stateful agent runtime | Yes. | https://omnia.altairalabs.ai/explanation/agents/memory/ | Yes. | documented | https://github.com/letta-ai/letta |
| Redaction before storage | Yes. *(enterprise)* | https://omnia.altairalabs.ai/how-to/privacy/configure-privacy-policies/ | Not documented. | not documented | https://docs.letta.com/configuration/memory/index.md |
| Consent and erasure | Consent revocation and a DSAR lifecycle. *(enterprise)* | https://omnia.altairalabs.ai/how-to/privacy/manage-user-consent/ | Per-item deletes. | partly documented | https://docs.letta.com/api-reference/agents/delete |
| Simulated users before release | Yes. | https://promptarena.altairalabs.ai/arena/explanation/scenario-design/ | Not documented. | not documented | https://docs.letta.com/agent-sdk/deployment/index.md |
| Stepped rollout, auto rollback | Yes. | https://omnia.altairalabs.ai/tutorials/progressive-rollouts/ | Not documented. | not documented | https://docs.letta.com/agent-sdk/deployment/index.md |
| Kubernetes | Helm. | https://omnia.altairalabs.ai/how-to/workspaces/manage-workspaces/ | Not documented. | not documented | https://docs.letta.com/self-hosting/index.md |

## Questions

**Is Letta open source?**

Its GitHub repository is Apache 2.0. Omnia is proprietary and free to run; PromptKit and PromptArena underneath are Apache 2.0.