Make the assumptions
work in the open.
Explore a monthly task scenario. All figures are user inputs or illustrative defaults; the model makes no prediction about agent reliability or realised savings.
Transparent calculation & limitations
Attempted tasks = volume × coverage. Baseline hours = volume × baseline minutes / 60. Retained hours = unattempted baseline hours + attempted review hours + fallback baseline hours. Capacity released = baseline hours − retained hours.
Net monthly value = capacity released × value per staff hour − attempted tasks × agent cost − monthly platform and operations cost. Simple payback = one-time implementation cost / positive net monthly value.
Every attempted task receives review; a fallback task additionally takes its full baseline handling time. This conservative overlap is deliberate. Negative capacity and negative value are retained. The model excludes tax, financing, quality effects, demand changes, ongoing implementation work beyond the entered operations cost, and any separately incurred fallback or retry model calls. Fold expected extra usage into the per-task cost. It does not estimate headcount reductions. Test assumptions in a controlled pilot.
Move from a scenario to a measured pilot
Collect actual baseline handling times and task volumes. Define completion quality and acceptable exceptions. Record review effort, fallback frequency, model usage and operating costs. Replace each illustrative default with measured evidence.
Prepare the governance checklist →