Built in, not bolted on.
Every action. Every decision. Every reason. Sealed. Powered by Lys — Ablative's immutable trace log, part of the Ablative Stack.
What it is
Lys is Ablative's immutable trace log. Every action taken by every AI agent in your system is logged in a tamperproof, cryptographically secured record. What was done. When it was done. Why it was done. What information informed the decision. That record can't be altered, deleted, or hidden — by anyone, including us.
This isn't a log file sitting on a server somewhere. It's a Merkle transparency log — the same cryptographic structure used by certificate authorities to secure the internet. Every entry is mathematically linked to the one before it, like a chain where each link is sealed to the ones before and after it. Change one record, and the entire chain breaks visibly. It's not tamper-resistant. It's tamper-evident.
Every decision is signed with Ed25519 cryptographic signatures — the same standard used in secure communications worldwide. Think of it as a digital wax seal: the signature proves who made the decision, when, and that the record hasn't been modified since.
Why "built in" matters
Most AI companies add audit trails as an afterthought — a logging layer bolted onto an existing system. Those logs can be incomplete, inconsistent, or quietly dropped when things go wrong. They exist to tick a compliance box, not to genuinely account for what happened.
Our trace log is different because it was never separate from the system. It's how the system works. The agents don't operate and then get logged — the logging IS the operation. Every decision, every recommendation, every piece of information accessed is part of the same immutable chain. You can't have one without the other.
No gaps.
Every action is captured because the system can't act without capturing it.
No tampering.
The log is cryptographically sealed. A record from three months ago reads exactly as it was written.
No ambiguity.
The full context of every decision is preserved — not just what happened, but why and based on what.
No complexity for you.
It's not a feature you need to configure or maintain. It's just there, working, from day one.
Among the first in the world
Ablative is among the first companies globally to ship user-facing decision provenance in a multi-agent AI system. That means our end users — business owners, team leaders, practitioners — can see exactly why the AI made each recommendation. Not just compliance teams. Not just auditors. The actual people receiving the recommendation.
Most companies that build audit trails build them for regulators. We built ours for you.
What this means for trust
Show why
When your AI recommends something, you can show why. Not "the AI said so." An actual chain of reasoning, backed by data, traceable to source.
Trustworthy records
When something goes wrong, the record can't have been modified after the fact. What it says happened is what happened.
Always ready
When a regulator asks, you don't need to scramble. The trace log has been running since day one. Everything is already there.
The bottom line
You shouldn't have to trust your AI. You should be able to verify it. Every action in our system is cryptographically logged in an immutable trace — tamperproof, complete, and available to you at any time. Not because regulation demands it. Because you deserve it.
Lys is open source (AGPL-3.0) and currently in development as part of the Ablative Stack.
AI you can verify.