A: BYOK (Bring Your Own Key) means we don’t bill you for the AI model tokens, but it doesn’t make the optimization pipeline unlimited or free on our side.

Even when the sitemap parsing, graph analysis, clustering, and scoring are executed by your AI provider via your key, there are still platform-level costs and constraints we have to manage:

• Job orchestration & infrastructure
Every optimization is still initiated, queued, monitored, retried, and finalized by our backend. We use serverless infrastructure (e.g. AWS Lambda and supporting services) to coordinate these workflows reliably, which carries real cost and ongoing maintenance.

• Concurrency & system stability
Unlimited, unthrottled jobs can easily overwhelm shared infrastructure during traffic spikes. Limits help us prevent bottlenecks and ensure consistent performance for all users — including those running large or complex sites.

• Pre- and post-processing work
Before and after the AI call, we still handle crawling coordination, data packaging, validation, visualization generation, storage, and report delivery. These steps consume compute, memory, bandwidth, and storage regardless of whose API key is used.

• Abuse prevention & safety controls
Reasonable limits protect against accidental loops, misconfigurations, or automated scripts that could degrade service quality or cause failures.

These limits are not set in stone. As we observe real-world usage patterns, optimize infrastructure, and gain confidence in performance at scale, we expect to increase them where it’s safe and sustainable.

BYOK gives you transparency and control over AI spend. The limits exist to ensure reliability, fairness, and long-term stability of the platform.

Hope that clarifies it 👍