Hands-on AI training
Three things, never mixed: your own numbers, industry figures with their sources, and what Symerse actually measures in the room.
01Your numbers
The worksheet asks only for figures you already hold:
It works out what the status quo costs you. The effect of training is a value you set and label as your own assumption. We never fill it in for you.
02Industry figures
These are other people's figures, with the year and a link. None of them is a Symerse result. Use them to sense-check your own numbers, not to replace them.
| Figure | Source | Year | What it means for a cohort |
|---|---|---|---|
| Over 80% of process deviations in pharma are attributed to human error. | PwC Belgium, Reducing human error in the pharma quality environment | 2025 | Human error is a causal factor, not a root cause. Judgement is part of what goes wrong, and part of what training can reach. |
| A deviation typically costs EUR 22,000 to 48,000, and up to EUR 880,000 with product loss. | PwC Belgium, same report | 2025 | Put your own cost per deviation in the worksheet; use this range to check it. |
| Lonza cut human-error-related deviations and non-conformities by more than 40% across 13 sites with an error prevention system. | Bodmann et al., CHIMIA 70(9) | 2016 | Evidence that a structured programme can move the number. It is not a Symerse result. |
03What Symerse measures
Every exercise ends in a scored decision, so a cohort leaves a record of how people reasoned. We report it in Learnovate's PEAS terms: Plan, Explore, Analyze, Share.
P · Plan
Each pair commits to a plan before the AI's view is shown, and the plan is kept in their course record.
E · Explore
What a pair asks the AI experts, what they flag and what they test all count towards the score.
A · Analyze
The score splits into analytical depth, evidence quality, safety and quality, and decision quality.
S · Share
The trainer debriefs every exercise, and each person leaves with a course record they can show.
A course record page (fictional trainee).
Illustrative
Illustrations of how to read the worksheet. They contain no Symerse results.
A growth site
A site bringing up a new line needs operators who can make sound calls on AI-assisted planning and quality review from the first batch. The number that matters is time to readiness: your onboarding hours per role, times the people you are hiring.
A site managing down
A site with fewer experienced people needs the next ramp-up to go right first time. The number that matters is the deviations you expect as experience leaves, priced at your own cost per deviation.
Only figures reconciled against the course records appear here.
Cohorts
2
Full two-day cohorts of the flagship course, with NIBRT in Dublin, in June and September 2026.
Product delivery record, 4 November 2026. Pilots and demo days are not counted.
No. No course can certify EU AI Act compliance, and we do not claim to. The course builds the practical AI literacy that Article 4 of the Act asks organisations to ensure in their staff.
A 30-minute call with the engineers who built it. No slides.