# Every Gen AI architecture article, by topic.

> Every InfoMover article in one place, grouped by topic and newest first. Each one answers a question CTOs, delivery heads and architects ask about Gen AI: how to design the system, how to prove it works and how to run it in production. Each one links to the consulting service that helps you do it.

## Gen AI in Production

How to take a Gen AI idea from a promising demo to a system your teams can rely on.

- [When Is a Gen AI Proof of Concept Ready for Production? Five Gates to Check](https://infomovertech.com/articles/gen-ai-proof-of-concept-to-production.md): A Gen AI proof of concept is ready for production when it passes five gates: its quality is measured on an evaluation set agreed with the business, it answers from your data with your permissions, its failure modes are tested and contained, it can be traced and budgeted, and a named owner is accountable after launch. Published: 2026-10-27. Web page: https://infomovertech.com/articles/gen-ai-proof-of-concept-to-production
- [What a Poorly Built AI Agent Really Looks Like, and What It Can Cost Your Business](https://infomovertech.com/articles/poorly-built-ai-agents-business-risk.md): A poorly built AI agent looks fine in a demo but runs with too much access, trusts every input, loops without limits, acts without approval and leaves no trace. The business pays in wrong commitments to customers, data exposure, runaway cost and lost trust. Governance, least-privilege security, observability and evaluation turn the same agent into a controlled system. Published: 2026-10-27. Web page: https://infomovertech.com/articles/poorly-built-ai-agents-business-risk