Blog
AI and automation in practice
We write about running costs, data quality, error handling and who is responsible once a system is live. Every piece deals with a decision that has to be made before an implementation, or while it is being extended.
The topics come from questions that keep returning in conversations about specific processes.
Latest articles
The topics come from questions asked before an implementation and after a system goes live.
What an AI assistant costs once the pilot ends
The invoice from a pilot describes a different system from the one that will carry your full traffic. What makes up the cost, and how to work out the cost of handling a single case.
ReadRAG is not the whole story. When classic search works better
Vectors are fashionable, but in half of our projects a good full-text index and a handful of rules are enough.
Readn8n in production - what we learned after 30 implementations
Error handling, monitoring, retries, deployment. What holds up in production, and what falls apart under the first real load.
ReadHow to train a team when half of it wants nothing to do with AI
A short guide for managers. No grand plans, no slide decks, just the practices we used with our recent cohorts.
ReadContact
Discuss your own process
The articles show how we assess a process. In a conversation we apply that to the data, systems and exceptions found in your company.
- 01
Process
You describe a task that takes up your team's time today and recurs in a predictable way.
- 02
Data and result
We establish which sources can be used and how you will know that the result is correct.
- 03
Further assessment
We set out what has to be measured before the scope of the implementation and its running cost can be defined.