announcing sift’s $42M series b funding. read more.
Codify engineering limits as rules, run them on live telemetry and historical data, and open the investigation with one click. Astrolab and Reliable Robotics detect anomalies in real time on Sift.
Traditional monitoring stacks raise the alarm and leave the diagnosis for later. Dashboards flash red. Thresholds fire at odd hours. Engineers chase false positives with spreadsheets and SSH sessions. Sift catches anomalies the same way engineers think about them: rules written in engineering terms, running on the actual telemetry, with the panel, timestamp, and triggering rule attached the moment the finding lands.
.png)
Define limits directly from your ICD — voltage bounds, pointing-error thresholds, rolling drift alarms. Compose conditions across channels, state transitions, and derived values without touching a DSL or config file. If your engineer can describe it, they can write it.
Run any rule against live telemetry and historical test data in the same pass. Validated detection logic from the bench carries straight into mission ops — no rewrites, no drift between environments. What you prove in dev is what protects you in production.
When a rule fires, the finding includes the channel, the timestamp, the triggering condition, and a direct link into the plot — cursor parked on the exact moment. Triage starts with the full picture, not a bare alert. No pivot to a separate tool. No asking someone to pull the data.
.svg)
Astrolab's engineers codify review criteria as Sift rules, turning senior engineer judgment into automated checks that run against every rover test campaign. The Rules feature has saved $100k+ by surfacing anomalies that would otherwise have slipped through manual review and forced a rework cycle downstream.
Reliable Robotics runs Sift's rules engine on the same telemetry streams that feed their flight certification process, catching anomalies in real time during autonomous flight testing. The rules engineers wrote in dev run unchanged in the certification environment, preserving the institutional knowledge that carries the program through regulatory review.
Write rules the way engineers describe the hardware: thresholds, state transitions, derived channels, rolling statistics. No PromQL, no custom DSL, no stapling a log query to a metric query to approximate a physics condition.
Every rule evaluates live telemetry and historical runs through the same code path. Write a rule today and Sift runs it across every past test in the program. Regression coverage comes free with the rule.
Anomalies surface with panel, timestamp, rule, and metadata attached. One click opens the plot. One more opens the investigation. Detection, triage, and root cause share one surface.
