Insights
AI and GenAI, from demo to production
Most AI projects stall between a demo that works and a system that works on real data, for real users, at a cost the business accepts. This section covers that gap: choosing between RAG and fine-tuning, what agents cost to run, when you need an MCP server, testing before launch, and which EU AI Act rules apply when.
Start here
| If you are asking | Go to |
|---|---|
| Can AMT build this for us? | AI development |
| Our pilot works in a demo but not in production | AI pilot to production |
| We need AI engineers in our team | Hire AI engineers |
Articles in this section
Coming soon
RAG vs fine-tuning vs long context: which one your use case needs
A decision table built on what the data looks like, not on what is fashionable
Coming soon
What an AI agent costs to run, per task
How to estimate model, retrieval and human review cost before you build
Coming soon
MCP servers explained for enterprise teams
What the Model Context Protocol is, when you need a server, and what to secure
Coming soon
How to test an LLM application before it reaches users
The evaluation set, the pass bar and who signs it
Coming soon
The EU AI Act in 2026: what applies now and what was delayed
Dated, sourced, updated when the rules change