Appearance
Decision Guide: AI Tool vs AI Feature
Use this guide when you are unsure whether your work is personal experimentation, a prototype, or a product capability.
AI tool
An AI tool is something you use to help with work.
Examples:
- asking AI to rewrite non-sensitive text;
- asking AI to explain public documentation;
- using an approved AI assistant to brainstorm ideas;
- using AI to generate mock data;
- using AI to help create a draft product specification.
This is usually an Experiment when it uses approved company accounts and appropriate data.
AI prototype
An AI prototype is a draft version of a possible workflow, product, or internal tool.
Examples:
- a demo that answers questions using mock data;
- a workflow that searches approved non-sensitive information;
- a local demo that calls fake APIs;
- a proof of concept for an AI assistant;
- an early specification that Dev can help evaluate.
This is usually an Experiment or Guided Prototype.
AI feature or managed tool
An AI feature or managed tool is something that may become part of a real company product, service, workflow, or operational process.
Examples:
- an AI workflow that needs production system access;
- an AI tool that uses customer, employee, financial, security, or company-confidential data;
- an AI assistant that takes action through tools, scripts, APIs, or agents;
- an AI capability that affects customers, employees, money, security, or operations.
This should become a Production Candidate and start handoff.
Simple rule
If the work helps you explore an idea with safe data, it can usually stay experimental.
If the work needs production systems, sensitive data, or real-world impact, start handoff with Dev.