Nebulock

Everyone complains about their SIEM because the approach to context is always the same: ingest everything into a central store, in case someone needs it later.

It's an expensive way to store data that mostly sits unread.

Today we’re launching Nebulock Helix: query data without pre-ingestion at the exact moment a hunt or investigation needs it.

Read more about Helix and how security operations doesn’t need to center around a SIEM.

nebulock.io/blog/nebulock-…

Your detection rules remember what they do, but they don't remember why.

Today, Nebulock is open sourcing the Agentic Detection Engineering Framework (ADEF): a durable, git-controlled memory layer for detection rules.

Agents draft the journal entries and flag duplicate logic or schema drift. Engineers still approve every change. Never lose rule context again.

To get started, clone the repo to get the CLI, Skills, MCP, and more.

nebulock.io/blog/agentic-d…

Not every security problem needs an LLM, and treating them all like they do can get expensive - fast.

Staff ML Engineer Stephanie Kirmer outlines why Nebulock built shadow AI hunt capabilities based on a classical ML model.

Using gradient boosted trees, the model returns same answer every time, exposes the decision process, & costs zero tokens to run.

nebulock.io/blog/using-cla…

Writing a detection rule takes minutes. Making it trustworthy can take hours.

Our Lead Detection Engineer Jarrett Polcari on automating the manual toil of detection engineering, and where automation should and shouldn't step in:

nebulock.io/blog/automatin…

We've raised $25M in Series A funding, led by @FirstMarkCap, with participation from @BainCapVC, @DecibelVC, @ZettaVentures, and Step Function.

We’re building a hunt-first security analytics platform to help defenders tackle the hardest challenge: threats that don’t trip alerts 🔽