Why Substr8 Exists

The Problem

AI agents can perform work.

But today, we cannot prove what they did.

When an agent:

• Processes a claim
• Makes a decision
• Calls external tools
• Generates outputs

There’s no verifiable record. Just logs that can be modified, deleted, or faked.

Why This Matters

For Developers

“What did my agent actually do?”

Debugging agents is hard. You need to know exactly what happened, in what order, with what inputs and outputs.

For Enterprises

“Can we prove compliance?”

Regulated industries need audit trails. Today’s agent logs don’t meet the bar for regulatory evidence.

For Multi-Agent Systems

“Can I trust this other agent’s work?”

When agents collaborate, they need to verify each other’s outputs. Without proof, it’s just blind trust.

The Solution

Substr8 generates cryptographically verifiable receipts for every agent execution.

Agent runs
    ↓
Substr8 captures every event
    ↓
Events form a hash chain
    ↓
Chain is Merkle-rooted
    ↓
Proof is signed (Ed25519)
    ↓
RunProof generated

The result: a RunProof — a portable artifact that anyone can verify.

What RunProof Proves

The Analogy

One Line

“Agents with receipts.”

Get Started

pip install substr8-langgraph

Quickstart →