# AI Unit Economics Model

## 1. Scenario inputs

| Input | Conservative | Base | Upside | Source / assumption |
|---|---:|---:|---:|---|
| Eligible customers/users |  |  |  |  |
| Adoption / active rate |  |  |  |  |
| Tasks per active user |  |  |  |  |
| Qualified success rate |  |  |  |  |
| Input/output tokens |  |  |  |  |
| Model/tool calls per task |  |  |  |  |
| Retry / failure rate |  |  |  |  |
| Human review rate/minutes |  |  |  |  |
| Price / value metric |  |  |  |  |

## 2. Customer value

| Value source | Gross value | Adoption/realization | Attributable annual value |
|---|---:|---:|---:|
| Time/capacity |  |  |  |
| Error/rework |  |  |  |
| Revenue/retention |  |  |  |
| Risk/cycle time |  |  |  |

## 3. Cost per task

| Cost | Per attempted task | Per successful task | Fixed / variable | Scaling note |
|---|---:|---:|---|---|
| Model |  |  |  |  |
| Retrieval/storage/network |  |  |  |  |
| External tools/data |  |  |  |  |
| Retry/failed work |  |  |  |  |
| Human review |  |  |  |  |
| Support/operations |  |  |  |  |
| Evaluation/security/engineering allocation |  |  |  |  |

## 4. Revenue and contribution

```text
Revenue = billable units × realized price
Variable cost = attempted tasks × cost per attempted task
Contribution = revenue - variable cost - directly attributable service cost
Contribution margin = contribution / revenue
```

## 5. Sensitivity

Test model price, tokens, steps, retry, success, human review, adoption, churn, discount, and support burden. Identify the three variables that can reverse the investment decision.

## 6. Decision

- Pricing/packaging choice:
- Economic break-even:
- Cost ceiling and monitoring:
- Experiment required:
- Stop or renegotiate trigger: