Introducing Autopilot: Never Overpay for GPUs Again

What is Autopilot?

Autopilot is Pulserun's flagship feature. It's a fully automated cost optimization system for GPU workloads.

You configure your workload requirements (GPU type, VRAM, storage) and set a budget. Autopilot handles everything else: finding the cheapest instance, launching it, monitoring prices, migrating when savings are available, and recovering from preemptions.

How It Works

1. Configure Tell Autopilot what you need: minimum GPU specs, Docker image, storage requirements, and your budget cap.

2. Deploy Autopilot scans all providers and launches your workload on the cheapest available instance that meets your specs.

3. Optimize While your workload runs, Autopilot continuously monitors prices across all providers. When it detects savings greater than 15%, it seamlessly migrates your workload to the cheaper option.

Key Specs

• **Auto-migration threshold:** 15% cost savings
• **Spot recovery time:** Under 30 seconds detection, under 2 minutes full recovery
• **Budget enforcement:** Daily caps with automatic shutdown
• **Active launch providers:** Vast.ai, RunPod, Lambda
• **Deferred until post-launch criteria:** TensorDock

Get Started

Autopilot is available to all Pulserun users. Enable it when launching any instance from the dashboard or API.