All work
LOG-02Live2026
A vendor-neutral business case for AI agents
An open calculator that ties an AI agent to the P&L: containment ramp, repeat contacts, per-minute vs per-resolution pricing, payback.
- Context
- Public tool · runs in the browser
- Role
- Designed the model, built and shipped it
- Shows
- Translating agent performance into money a CFO will sign off on, and knowing where the numbers lie.
- 17
- inputs, every one stated and adjustable
- 2
- pricing models compared side by side
- c*
- the containment rate where the two pricing models cost the same
Problem
Most AI-agent business cases are a vendor slide: a containment number, a cost per minute, a savings line that goes up and to the right. They skip the three things that decide whether the project actually pays back.
What the model gets right
- Containment ramps. Go-live containment is not steady state. The model ramps it linearly over the months you choose.
- Containment is not resolution. A share of "contained" contacts call back and reach a human anyway. They cost you twice.
- Pricing changes the answer. Per-minute pricing charges for every AI minute, including the ones spent before a handoff. Per-resolution pricing charges only for contained contacts. The calculator computes the containment rate at which the two cost the same: below it per-resolution is cheaper, above it per-minute wins.
Using it in a deployment
Run it in discovery with the customer's own numbers. The output is a two-page summary (print or save as PDF) with every assumption stated, so finance argues with the inputs, not with you.
What it deliberately does not model
CSAT and revenue effects, attrition and hiring, peak staffing (Erlang), compliance cost. They matter. They are also where business cases get inflated, so they stay out of the headline number.