Onboarding flow

Seven guided steps on first launch — profile, hardware, benchmark, models, preferences, and node registration.

When it runs

Onboarding triggers exactly once on the first app launch after install. A local flag prevents it from showing again after completion. If the app closes mid-flow, it resumes from the last completed step.

Steps are sequential — you can go back within completed steps but cannot skip ahead until required data is valid.

The seven steps

Step 1

Welcome

Full-screen intro with optional welcome video. Explains earning by sharing GPU power. Runs once; stored in local app storage.

Step 2

Provider profile

Collect display name (public to developers), timezone (auto-detected, overridable), and optional payout address. Synced to the gateway when the node registers.

Step 3

Machine detection

Automatic scan: GPU model & VRAM, CPU cores/threads, RAM, disk space, OS version, network speed test. Critical items (GPU, CPU, RAM) must succeed before continuing.

Detailed guide →

Step 4

Benchmark

Inference benchmark on your GPU: tokens/sec, time-to-first-token, peak VRAM, utilization, temperature. Produces a normalized score (0–10000) used for routing and earnings projection.

Detailed guide →

Step 5

Model recommendations

Catalog fetched from the API gateway, filtered by your VRAM and benchmark score. Queue compatible models for download.

Detailed guide →

Step 6

Contribution preferences

Monthly earnings target, weekly availability hours, preferred workload types (text, image, embeddings, etc.), and data sensitivity level. Drives routing filters and earnings preview.

Step 7

Earnings preview & register

Live projected earnings from benchmark + preferences + gateway market rates. Register the node to receive a node ID and start earning.

Detailed guide →

Data collected

All onboarding data is persisted locally first, then sent to the API gateway on registration:

  • Provider profile (name, timezone, payout address)
  • Hardware specification snapshot
  • Benchmark results and score
  • Contribution preferences
  • Returned node identifier after successful registration

After onboarding

You are not live yet. Download and select models, run preflight, then go live from the main dashboard.

Re-run onboarding

To update hardware or preferences later, use Settings in the desktop app or re-run benchmark from the provider web dashboard at /provider/benchmark. The full wizard does not repeat automatically.

Next