Updates, thinking, and technical writing from the DarkMatter team.
March 2026
announcement
Introducing DarkMatter: execution lineage for AI agents
Multi-agent AI systems are moving fast. What's missing is the layer underneath — the one that remembers exactly what happened, lets you replay any decision, and proves the chain was never tampered with. That's what we built.
Read post →
Coming soon
technical
Why we used SHA-256 hash chaining — and what it means for audit trails
How DarkMatter's cryptographic chain works, why parent hash chaining matters more than simple checksums, and what "tamper-evident" actually means in practice.
Coming soon
use case
Fork as a first-class primitive: branching AI workflows like Git branches code
What it means to fork an agent execution at a checkpoint, why this is qualitatively different from retrying a prompt, and how teams use it to explore alternate reasoning paths.