World-class agentic infrastructure, built for serious users and businesses.
Two products, one foundation. Memory that's proven, not guessed. Compliance that's enforced, not promised.
Sibyl Memory
Live
Long-term memory for AI agents.
Most AI forgets everything the moment a chat ends. Sibyl Memory gives it a memory that lasts: the people, facts, and decisions that carry from one session to the next. It placed #2 on LongMemEval (95.6%), a leading benchmark for long-term AI memory, and it is the only file-based system in the top tier.
Install in one line. Works with Claude Code, Cursor, Codex, and Hermes Agent. Free to start, no account needed.
Simple pricing for Sibyl Memory.
Start free. Upgrade when you need more. No lock-in, cancel anytime.
For one person, on one device.
- The full memory plugin, running on your machine
- One device, one user
- Comes with a storage limit
- Free forever, no account needed
For power users, across all your devices.
- Everything in Free
- No storage limit
- Syncs across all your devices
- Self-learning: your memory sharpens itself as you use it
For teams and multi-agent setups.
- Everything in Pro
- Shared memory across people and agents
- Multi-agent fleet management
Hold $SIBYL? Stakers get a paid tier free.
Sibyl Sovereign
In developmentGoverned agents your business can be accountable for.
Sovereign builds AI agents businesses can trust with real duties: paying invoices, handling customer records, sending communications. Handing an AI real power is risky, because it can make mistakes and it can be tricked. So Sovereign puts a deterministic guard between the agent and the real world. The agent proposes an action; ordinary, predictable software decides whether to allow it. That guard uses no AI, doesn't guess, and can't be talked out of its rules.
The agent can be wrong.
The gate can't.
-
Identity, purpose, and goals
A plain-English charter that sets who the agent is, what it's for, what it may touch, and its limits. Written so a compliance officer can read it, nothing hidden in code. -
Security
A cryptographic seal on every critical file, re-checked before every action. Tamper-evident, fail-closed, and isolated to your data only. -
Compliance, enforced
A deterministic gate checks every action against your rules before it runs. A payment over the limit, to an unapproved address, or outside policy simply never executes. -
Provable
Every decision is written to an append-only audit trail. No record, no action. What we enforce, we prove; what we can't force, we monitor and surface.
We don't claim the agent never errs. We make sure its errors can't reach the world, and we show you everything we can't guarantee rather than hide it.
In development, and being proven against real business workflows like accounts payable and accounts receivable. If you're deploying agents under a compliance mandate, we're taking early conversations.
Building something serious? Let's talk.
Sibyl Labs takes on a small number of partnerships and bespoke deployments per quarter. We do not do volume.