Continuous Monitoring

Continuous Monitoring is the recurring counterpart to AI Result Evaluation. Instead of requesting an evaluation once, you define a schedule — a source (verification or Language Technology runs), a cadence (hourly, daily, or weekly), and an optional project scope — and Crowdee periodically samples a recent run, dispatches a crowd evaluation for it, and tracks the resulting transparency score over time.

Each scored run is compared against that schedule's own trailing average of prior completed runs, and flagged the moment it deviates meaningfully — there's no absolute quality bar to configure, since every schedule calibrates against its own history.

What You Get

• A recurring evaluation, without recurring manual requests. Every tick dispatches a normal AI Result Evaluation batch under the hood.
• Self-baselining drift detection. Each schedule flags deviation from its own history, not a fixed threshold you have to tune.
• A full run history, including empty periods (no eligible run to sample) and failures (for example, insufficient credits), so you can audit coverage over time.

Cost

Empty ticks (nothing eligible to sample) are free. A tick that dispatches an evaluation costs the same as one AI Result Evaluation — 45 credits by default — blocked from your organisation balance at dispatch time. There's no separate subscription fee for the schedule itself.

Where to Go Next

See Continuous AI Monitoring for the full technical reference: schedule fields, tick behavior, run statuses, and the drift-detection formula.