we refuse to
build another
dashboard.
The observability industry sells you visualizations of your own ignorance. Colorful graphs that move when things break. Alerts that fire after your users already tweeted about the outage. We are done pretending this is acceptable.
null.systems is developer infrastructure built on a single conviction: your tools should understand your systems better than you do. Not approximate. Not sampled. Not eventually consistent. Precise, causal, real-time understanding.
We are building the last observability tool you will ever install. It is open source. It is opinionated. It does not have a dark mode toggle because it does not have a UI at all.
observability is broken.
The average engineering team spends 14 hours per week context-switching between monitoring tools. They maintain 6+ dashboards per service. They still get paged at 3am for issues their tools should have predicted 45 minutes earlier.
Current solutions ingest terabytes of telemetry and charge you per gigabyte to store data you will never query. They give you 200 integrations when you need 12. They optimize for demo day, not day two.
The result: $40B spent annually on tools that make engineers feel busy instead of informed. We think that is a market failure worth fixing.
infrastructure that thinks.
null.systems is a single binary that replaces your metrics pipeline, log aggregator, and APM vendor. It runs in your cluster, keeps data in your network, and uses causal inference to tell you why something broke before you ask.
ZERO SAMPLING
Every event. Every span. Every log line. We compress at the edge using semantic deduplication, not random sampling. You keep 100% fidelity at 10% of the storage cost.
CAUSAL GRAPHS
We model your infrastructure as a directed acyclic graph of causal relationships. When a node fails, we trace the blast radius in milliseconds and tell you the root cause, not the symptom.
CLI NATIVE
Query your infrastructure like you query a database. Pipe results. Script automations. Our query language is designed for operators who live in the terminal, not managers who live in slides.
four steps. no magic.
INSTALL THE CLI
One command. No GUI wizards. No account creation. Pull the binary, authenticate with your SSH key, and you are running in under 30 seconds.
POINT AT YOUR STACK
null.systems reads your infrastructure declaratively. Kubernetes, Terraform, raw VMs, it does not matter. We parse your topology and build a real-time dependency graph.
OBSERVE EVERYTHING
Every packet, every syscall, every deployment. We instrument at the kernel level so you never miss a signal. Zero sampling. Full fidelity.
ACT ON TRUTH
Alerts that fire on actual anomalies, not threshold breaches. Our detection engine uses causal inference, not pattern matching. False positive rate under 0.3%.
open by default.
The core engine is Apache 2.0 licensed. Always has been, always will be. We believe infrastructure software that holds your data hostage is a liability, not a product.
Our commercial offering adds multi-tenant management, SSO, audit logging, and 24/7 on-call support. The detection engine, query language, and ingestion pipeline are free and open for anyone to run, fork, and improve.
the people.
KIRA VOSS
CEO / SYSTEMSEx-kernel maintainer. Built observability at Cloudflare for 6 years. Believes dashboards are a coping mechanism.
OMAR ZHANG
CTO / INFRAFounded two open-source projects with 40k+ stars. Previously principal engineer at Datadog. Writes Rust for fun.
SABLE OKAFOR
HEAD OF PRODUCTShipped developer tools at Stripe and Vercel. Obsessed with reducing time-to-insight below 200ms.
join the waitlist.
We are onboarding teams in batches of 20. Priority goes to teams running Kubernetes in production with more than 50 services. No sales calls. No demos. Just access.