Why most AI pilots stall before production
The gap between a working demo and a production system is rarely the model. It is permissions, error handling, evaluation, cost control, and the ordinary conditions of a working day.
Insights
Field notes from building things that have to work.
Practical notes on AI implementation, automation, digital products, and growth, written from delivery experience rather than from a content calendar.
- Written by the people doing the delivery
- No vendor marketing, no invented statistics
- Short, practical, and specific
- Published when we have something worth saying
What an AI receptionist should never do
The value of a voice agent is in what it refuses to attempt. Boundaries are the product, not a limitation of it.
Retrieval quality beats model choice
Teams spend weeks comparing models and an afternoon on how documents are prepared. The second decision usually determines whether the answers are any good.
Automate the exception, not just the happy path
Most of the cost in a manual process is not the standard case. It is the twenty percent that does not fit, and that is usually the part automation projects leave behind.
Your website is an operations problem
A site that generates enquiries nobody can service on time is not a marketing success. It is a queue with better typography.
Report on enquiries, not clicks
Impressions, clicks, and sessions are inputs. If a report stops there, it is describing activity rather than outcome, and it will survive far longer than it deserves to.
NEXT STEP
Have a question we have not written about?
Send it with your brief. If it is a good question, the answer usually becomes the next note published here.
We answer enquiries within one business day.