MultiTool docs

MultiTool is a progressive delivery tool that catches production issues before they impact users. MultiTool's open-source CLI connects with the MultiTool app to manage canary deployments in real time. As changes are rolled out, MultiTool monitors key metrics and automatically shifts traffic, promotes stable releases, or rolls back changes if regressions are detected.

The problem MultiTool solves

Deployments can fail, and when they do, users are often the first to notice. For high-stakes releases, engineers actively watch dashboards and verify system health. But for most routine deployments, this level of attention isn't feasible. Instead, teams rely on alerting systems to detect issues after they've already surfaced as a problem in production. Both approaches have drawbacks: actively monitoring dashboards is labor-intensive and doesn’t scale, while waiting for alerts is reactive and slow. In both cases, the failure is detected too late.

MultiTool takes a different approach. It continuously analyzes service behavior during a rollout and builds a statistical model to decide whether a deployment is healthy. Based on this evaluation, it can automatically promote stable releases, pause suspicious ones, or roll back regressions without requiring human intervention.

A proactive model helps teams deploy more confidently and more frequently, while reducing operational overhead. It also eliminates the need for internal tooling to manage traffic shifting, rollback logic, or rollout monitoring.

How MultiTool works

MultiTool manages progressive rollouts by using controlled canary deployments. When a new version of a service is deployed, only a portion of traffic is initially directed to the new version (the canary), while the rest continues to flow to the stable baseline. MultiTool monitors key metrics like error rate, latency, and CPU usage, then builds a performance benchmark based on previous healthy releases. The canary is compared against this benchmark in real time. If its behavior deviates significantly, MultiTool treats it as a regression and rolls it back. If the canary performs within acceptable limits, traffic gradually shifts until the rollout is complete.

MultiTool is composed of two components:

• An open-source CLI that runs in your environment. It initiates deployments, reports service metrics, and adjusts traffic between the baseline and canary versions.

• A hosted SaaS application that manages rollout logic. It builds and maintains the performance model, determines when to shift traffic, and decides whether to promote or revert a deployment based on observed behavior.

Together, these components enable safe, automated rollouts that respond to live service conditions without manual oversight.