A vendor-selection workbook for AI phone agents: pricing decoder, 40-point scorecard, demo-call script, 14 adversarial test calls, contract checklist, and a 30-day pilot plan.
An AI receptionist is the first thing a stranger experiences about your business, running unsupervised, at 7:40 PM on a Saturday. Most owners buy one off a demo and a discount, then switch vendors twice in eighteen months. This workbook exists so you do not. It starts by making you put a number on your own missed calls — call volume, after-hours split, close rate, average job value — because every vendor's ROI slide uses their number and you need yours before you take a single demo. From there it is a working comparison process. You get the four vendor archetypes (horizontal AI answering services, vertical industry platforms, human-plus-AI hybrids, and builder platforms) with published September 2026 pricing for each, so you can see why a $49/month product and a $2,100/month product are not competing for the same job. You get a pricing-model decoder that explains why per-minute, per-call, per-unique-caller, and component-based billing produce wildly different bills on the same call volume — plus the eleven line items vendors do not lead with, from concurrency lines to HIPAA add-ons to whether spam calls are metered. You get a 40-point weighted scorecard across call handling, integrations, escalation behavior, configurability, reporting, compliance, commercial terms, and support, with explicit disqualifiers that end an evaluation regardless of total score. The two chapters buyers report using most are the demo script and the test-call battery. The script is 22 questions written to be hard for a salesperson to dodge, including the one that reveals the most: why do customers like us churn? The battery is 14 calls you make yourself — price fishing, an unavailable slot, an emergency, an angry caller demanding a human, three simultaneous calls at 2:47 AM — designed to surface confident-wrong behavior rather than confirming that the voice sounds nice. The guide closes with a nine-term contract checklist, a 30-day pilot plan with a real day-30 decision gate, the nine metrics worth tracking afterward, and an honest chapter on the five situations where the answer is that you should not buy one yet.
Every one of these is free and email-gated, same as this one.
How to choose an AI-assisted accounts payable, spend, and close stack: category map, pricing decoder, scorecard, six-document torture test, fraud controls, and a six-week migration plan.
PDFDecode per-resolution pricing, audit your knowledge base, and run a real 200-question bake-off before you sign an AI customer support contract.
PDFCapture every lead, answer every call, and keep past clients buying and referring on autopilot — without hiring a full-time ISA.
PDF