2026-04-23 · 4 min read · GO Pilot GO Team
Most agent platforms charge per run. We charge per resource consumed: tokens, voice seconds, and browser seconds. Three reasons.
First, a "run" is a meaningless unit. A run that summarizes one email and a run that scrapes 40 pages and drafts a proposal are not comparable. Run-based pricing penalizes short useful automations and subsidizes runaway ones.
Second, resource metering matches our actual cost structure. We pay LLM providers per token, ElevenLabs per second of voice, and Browserbase per minute of headless browser session. Passing through real cost lines, rather than averaging them into a "run fee," means you only pay for what your agents actually do.
Third, it makes optimization legible. If you see your token spend climbing, you can shorten system prompts, trim memory, or switch a step to a cheaper model. If your browser minutes are climbing, you know to add caching or narrow your scrape targets. With run-based pricing those signals are invisible.
Our plans include generous baseline quotas: Starter ships with 5M tokens, 60 voice minutes, and 30 browser minutes per month. Pro ships with 25M tokens, 300 voice minutes, and 180 browser minutes. Add-on packs are available at cost-plus when you need more, and a daily cost ceiling per workspace acts as a hard kill switch so a runaway loop cannot drain your card.