AI ROI & Unit Economics Calculator

Load a scenario

Inputs

Scale

Monthly Active Users (MAU)users/month

AI tasks per user per monthsessions or requests

Task completion rate% tasks AI handles without escalation

Model costs (per task)

Avg input tokens per taskprompt + context

Avg output tokens per taskgenerated response

Model price — input ($ per 1M tokens)

Model price — output ($ per 1M tokens)

RAG / retrieval overhead ($ per task)vector search, reranking; 0 if none

Avg retries / fallbacks per task1 = no retries

Value generated

Human cost per task (if not deflected)$ — agent time, support cost, etc.

Revenue uplift per completed task$ — conversion, upsell; 0 if cost-saving only

Business

Target gross margin %floor you need to clear

Results — monthly

Scale

Total tasks/month —

AI-handled tasks —

Escalated to human —

Cost breakdown (monthly)

Model inference cost —

RAG / retrieval cost —

Total AI cost —

        Cost per completed task
        —
      

Value generated (monthly)

Human cost deflected —

Revenue uplift —

Total value —

Economics

Net monthly impact —

Gross margin on AI work —

How to use this: Start with a preset, then tune to your actual model pricing and product assumptions. Prices vary — check your provider's current pricing. “Human cost per task” is the fully-loaded cost of a human handling that task (agent time, management overhead). “Cost per completed task” is the number to watch as usage scales — see Unit Economics §7 for the full framing.

What this doesn't include: One-time training/fine-tuning costs, infrastructure fixed costs, monitoring and observability tooling, human-review overhead for the escalated tasks. Add those to your cost model before final go/no-go.