# When to sell an AI agent service and when to keep it internal

A margin-first way to decide whether an agent workflow belongs in a client offer or should stay inside operations.

- Section: AI Agent Revenue
- Content type: Money Guide
- Truth label: analysis
- Updated: 2026-07-01
- Source count: 0
- Confidence: medium
- Disclosure: testing: desk researched

## Agent brief

A decision guide for choosing whether an AI agent workflow is ready for a client-facing offer.

Source basis: Desk framework fixture for validating homepage revenue slot behavior.

## Commercial takeaway

- Who should care: consultants, agencies, solo operators.
- Commercial use: Helps operators avoid selling workflows before the delivery margin is credible.
- Cost basis: Requires existing agent tools, review time, and client-specific setup work.
- Proof standard: hypothesis.

## Why operators should care

Helps operators avoid selling workflows before the delivery margin is credible.

## Commercial path

Money claim type: make money.
Package only workflows with a reviewed deliverable, clear savings, and documented human oversight.
Estimated cost: Requires existing agent tools, review time, and client-specific setup work.

- Do not sell agent output as autonomous delivery without review evidence.
- Client data access, revision time, and support needs can erase apparent margin.

## Checks and risks

- Risk: unsupported revenue framing
- Risk: hidden review labour
- Risk: client data access limits

## Source basis

- No external sources listed. Source basis: Desk framework fixture for validating homepage revenue slot behavior.

## Related links

- [Can an AI agent actually make money? Start with the operator margin](/ai-agent-revenue/operator-margin-ai-agents/)

## Machine-readable

- Read: /ai-agent-revenue/agent-service-margin-check/
- Article JSON: /ai-agent-revenue/agent-service-margin-check/article.json
- Brief JSON: /ai-agent-revenue/agent-service-margin-check/brief.json