01 / USER
Task completion
Can a representative user complete the intended journey, and is the outcome genuinely useful?
PRACTICAL PRODUCT PLANNING
Define the smallest useful product, its data boundaries, acceptance criteria and operational responsibilities before development.

Practical tools. Clear advice. Real decisions.
DEFINE BEFORE YOU BUILD
A minimum viable product is not simply an incomplete full application. It is a carefully limited version that lets a defined user complete an important task and provides evidence about whether the proposed solution works.
Before choosing an AI builder or developer, capture the intended user, triggering problem, the expected output and how the result will be judged. This avoids mistaking a functional-looking prototype for an acceptable product.
ACCEPTANCE CRITERIA
01 / USER
Can a representative user complete the intended journey, and is the outcome genuinely useful?
02 / TRUST
Are calculations, guidance and data flows correct? What happens when inputs are missing or misleading?
03 / OPERATION
Who maintains the content, handles errors, manages backups and pays for the service over time?
Yes. Generative tools can speed up implementation but do not automatically validate authorisation rules, secure storage, dependencies, observability, data protection or business accuracy. Human review and testing remain part of a credible delivery process.
Document the difference between a prototype and the deployed operating model: performance, incident response, releases, security patches, monitoring, accessibility, service ownership and exit options. Plan that transition rather than assuming it arrives with the prototype.
For data-location decisions see EU data residency: what needs checking? and the AI-built app security checklist. For earlier product ideation see finding app ideas in existing content.
Describe its users, expected outcome and any sensitive information it needs.
Discuss requirements