No engineer has time to scroll through a long PDF of technical information to search for alarm codes. That is why claims about AI saving time need clear evidence behind them. In contrast, Machinery Market have set out the Machinery Market’s account of the AMRC Cymru and Instro AI trials in detail, for us to check as we read.
Designing the trials this way is far more valuable than any single percentage figure. Good to see Poeton Industries reporting specific figures rather than meaningless excitement.
A percentage means little without a baseline
This means that instead of simply reading that they have improved in solving technical information queries by two-thirds, Colchester Machine Tool Solutions have actually improved from locating and responding to technical information within 5.5 minutes to 1.8 minutes, a massive 67.3% improvement. Timed tasks are so much easier to compare than satisfaction scores.
Defined objectives are the first test

The feature highlights that the programs were developed to achieve specific business objectives. When testing features that use AI, it’s crucial to define what success looks like prior to commencing the test. The Colchester engineers ran representative service and maintenance tasks for CNC and manual combination lathes in the test. These are real tasks, not demo queries that have been set up to provide the best results.
Joint validation adds weight
Joint validation by AMRC Cymru is more influential than allowing a supplier to report their best week.
Two yardsticks at Poeton Industries
As Lee Mason, group digital transformation manager at Poeton Industries, comments, Phase 2 of the trial will scale that progress, deepen the use cases and test how the tool embeds into daily operations.
Good trial design earns the numbers their weight
This means that any company in the surface treatment industry can read a supplier’s claim and assess it themselves by asking for a number to start with, asking whether real customer inquiries or artificial ones were used in the test and who has checked the results. In Poeton’s case, two different measurements have been determined during the trial and thus the results have a solid basis.

