Job task analysis for utility training: when a procedure changes, the tasks, questions and retrain list change with it, before anyone switches the old way
Training built from the procedures crews actually use.
See one case, screen by screen ↓A procedure revision arrives as a PDF with change bars. Someone in the training team has to read it against the task analysis, find each step that came from a changed clause, find each question that tests that step and each slide that teaches it, re-rate the task, then work out from the authorisation register who holds it and when they next switch. The learning system records who completed what, not which clause a lesson came from, so all of that is done by hand from a spreadsheet and memory.
Until it is done, the question bank still passes people who give the answer the old procedure wanted. At the demo utility, the rev 7 question “At the RMU, how do you confirm you have the right panel?” still accepted “check the panel label against the schedule”, the check that was done in the wrong-panel near-miss. For a reliability-related task, NERC PER-005-2 R3.1 asks for each person's capabilities to be verified within six months of the change, and R1.1 asks for the task list to be built by a documented method and reviewed each calendar year.
Estimated split for one procedure revision that changes three tasks, in working days, by hand and with the solution. The by-hand total matches the median of 7 weeks in the demo utility's 2025 change log (14 revisions, 6–10 weeks each).
Six specialist agents turn each procedure revision into changed steps, redrafted questions and a retrain list; a subject-matter expert signs every item.
Lumina Utilities trains its HV authorised persons at Lumina Academy's Fernhollow Training Centre. The job “HV authorised person: distribution switching” has 31 tasks, taught by 9 lessons and tested by 310 questions. On Friday 9 October 2026 its 11 kV ring main unit procedure changed. Here is what happens next, in the working solution.
Joanne issues OP-SW-214 rev 8, “Switching and isolation of 11 kV ring main units”, under change notice CN-2026-041. It follows near-miss NM-2026-118 at Kilnsey Lane 4 RMU and lesson learned LL-2026-07 from the Fernhollow rig. The change inbox picks it up as a task change, beside a chlorine drum changeover revision at Saltmere water treatment works, a new RMU type commissioned at Ashgrove and a wording-only change to the LV pillar fuse procedure.
NM-2026-118: “Hand on panel 2 when the schedule said panel 3; stopped by the second person”
The procedure reader compares rev 8 with rev 7 clause by clause. §5.4 is new: lock the local/remote selector in LOCAL with a safety lock and record the lock number on the switching schedule. §6.3 replaces “check the panel label” with point and confirm. §7.2 adds a second proving of the voltage indicator after testing for dead. By 14:25 the task analyst has updated HV-B12, HV-B07 and HV-C09: 4 steps changed and 2 added, each cited to its clause.
“Procedure reader · §5.4 new · §6.3, §7.2 changed”
Amira owns the task analyses and assessments for HV switching and earthing. She opens the breakdown: the procedure on the left with change bars, the tasks in the middle with each step tagged to its clause, and the questions on the right with the step each one tests. The banner reads 3 clauses changed, 3 tasks, 7 questions redrafted or added, 64 people retrain, 11 this week.
“Draft · not live until an SME signs”
One click shows rev 7, the training people are assessed on today. HV-B12 stops at telling the control engineer the RMU is in local control: there is no selector lock and no lock number. Its question “You have selected LOCAL. What do you tell the control engineer?” and the HV-B07 question “At the RMU, how do you confirm you have the right panel?” are marked as testing steps that have changed.
“Showing rev 7: the training that is live today”
The task statement reads as given, do and to the standard: transfer the RMU to local control and secure the selector so that it cannot be operated from the control room while anyone works on its circuits. Importance goes from 3 to 4, because in IR-2025-044 a selector left in REMOTE let the control room close a switch during cable work and a jointer was burned. Difficulty goes from 2 to 3 for the lock and its record. Rhys Calloway and Megan Holt agree; Yusuf Haddad has not yet confirmed.
The criticality rater only proposes: three SMEs confirm a critical rating.
Q-B07-3 now asks: “You are at Kilnsey Lane 4 RMU to open panel 3. What must happen before you operate it?” The right answer is point and confirm, cited to §6.3. Distractor A is the rev 7 answer, the check that was done in NM-2026-118; distractor C, asking the second person, is what stopped the near-miss but is not the rule. The rev 7 item is retired when the new one is signed, never deleted.
Rhys Calloway: “If a question can be passed by someone who'd still open the wrong panel, it's the wrong question.”
L-HV-214 “Point and confirm, double proving” is drafted from the changed steps: why it changed, point and confirm, locking the selector, prove-test-prove and a knowledge check, then the practical on the Fernhollow 11 kV RMU rig with instructor Colm Brennan. Each objective maps to a step and the question that tests it. Six of 22 slides in deck HV-SW-04 must change, and the practical checklist marks five critical steps.
Practical checklist: “observed on the rig · critical steps 100%”
From the authorisation register and the switching schedules, the impact mapper finds 64 authorised persons in four depots: Brackenford 19, Kilnsey Cross 16, Marrow Bay 17 and Ashgrove 12. Eleven are named on switching schedules for these tasks between Wed 14 and Tue 20 Oct, so they get the 45-minute briefing first. The first is Gareth Lowe, on Wed 14 Oct at 09:00, at Kilnsey Lane 4: the near-miss site.
The impact mapper never suspends an authorisation; the training manager decides.
The practical re-assessment is due six months from the revision, by 9 Apr 2027. The plan proposes 8 sessions of 8 seats on the Fernhollow rig, from RA-01 on Tue 3 Nov at Brackenford to RA-08 on Thu 18 Mar as a mixed catch-up session, with Colm Brennan as instructor and Rhys Calloway as assessor.
“8 sessions × 8 seats = 64”
Daniel signs the retrain plan with his password: 11 briefings before the next switch, and a practical re-assessment for all 64 authorised persons in 8 sessions by 9 Apr 2027. Nothing from rev 8 reaches a person until three people have signed: Rhys for each changed step and question, Amira for the three task analyses, and Daniel for the plan.
Daniel Okoro: “Tell me who must be re-briefed before they next switch, not after.”
The same revision, seen by the four people who own it: what a procedure change meant before, and what it means now.
Six specialist agents turn each procedure revision into changed steps, redrafted questions and a retrain list; a subject-matter expert signs every item.
Compares each procedure revision with the issued version, clause by clause, quotes the old and new wording and classifies every change as a task change, a new task or wording only. Links the incidents and lessons learned the change notice cites.
Breaks a procedure into tasks and steps, writes each task statement as “Given …, do …, to the standard …” in observable terms, lists the knowledge and skills each step needs and tags every step with its source clause.
Proposes difficulty, importance and frequency, each 0–4, and the training decision, using incident history for importance and switching-schedule counts for frequency. A score of 10 or more, or importance 4, is critical.
Redrafts or adds knowledge questions, scenarios on named equipment and observed practical checks for each changed step, with distractors drawn from the old procedure and from real errors in incidents, and the clause behind the right answer.
Updates the lesson plan for the changed steps: objectives mapped to steps, timing, activities, the rig practical and its checklist, and every slide that still teaches an old step.
Lists everyone in the authorisation register who holds an affected task, reads the next 7 days of switching schedules so anyone named is briefed first, sets the practical re-assessment six months from the revision and proposes rig sessions of 8.
The expert signs each item; the manager approves who retrains. The agents propose; a named person decides.
Ask in plain words, or say what to change. Every answer points to the clause, schedule or record behind it.
What changed in rev 8, and what does it break?
Three clauses, three tasks, seven questions. §5.4 is new: lock the local/remote selector and record the lock number. §6.3 replaces “check the label” with point and confirm. §7.2 adds a second proving of the indicator. That changes 4 steps and adds 2 in HV-B12, HV-B07 and HV-C09; 5 questions tested the old steps and 2 new steps had none.
Who must be briefed before they next switch?
11 of the 64. Their switching schedules fall between Wed 14 and Tue 20 Oct, starting with Gareth Lowe at Kilnsey Lane 4 tomorrow at 09:00, the RMU where the near-miss happened. The other 53 have no switching on these tasks this week; all 64 need the practical by 9 Apr 2027.
Why is HV-B12 now critical?
Its importance went from 3 to 4. A selector left in REMOTE let the control room close a switch during cable work at Marrow Bay, and a jointer was burned. Rev 8 adds a lock and a record, so difficulty rises from 2 to 3. The score goes from 7 to 9, and importance 4 makes any task critical. Rhys and two other SMEs still have to confirm it.
Draft the point-and-confirm question in Arabic too
Done, as a draft beside the English. The four options keep the same order and the same answer. It goes to Rhys with the English item; an Arabic-speaking SME, Yusuf Haddad, is added as a second reviewer.
13 screens from the working solution, on its sample data. Pick one to see it large.
How tasks are rated, when people retrain, who signs and where the agents read from are settings, saved as a new version each time they change.
| Setting | Default | Choose from |
|---|---|---|
| Critical at a score of | 10 | difficulty + importance + frequency, each 0–4 |
| Importance 4 is always critical | On | On · Off — importance 4 means injury or loss of supply is possible |
| SMEs to confirm a critical rating | 3 | a number of SMEs — the rater only proposes |
| Use incident history for importance | On | On · Off — incidents and near-misses since 2019 |
| Critical task changed: briefing before the next use | On | On · Off |
| Practical re-assessment within | 6 months | months from the revision |
| Knowledge check pass mark | 80 % | per cent; critical steps on the practical must all be passed (100 %) |
| Look ahead in switching schedules | 7 days | days — anyone named in that window is briefed first |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate; 6–10 weeks and the working-days split are estimated from the demo utility's 2025 change log · “typical” = published figures (BLS Census of Fatal Occupational Injuries via Relia Magazine: line installers 13.8 deaths per 100,000 against 3.3 for all workers, not re-checked). Regulatory context: NERC PER-005-2 R1.1 and R3.1; DOE-HDBK-1078-94, A Systematic Approach to Training. Lumina Utilities, Lumina Academy and the people, depots, equipment, procedures and records named on this page are fictional — characters and sample data in the working solution.
Software that turns a utility's procedures into a task list with steps, knowledge, skills and criticality, and into the lessons and assessment questions built on them. Job Task Analysis keeps every step and question joined to the procedure clause it came from, so when a procedure changes, the training that depends on it can be found and updated.
The procedure reader compares the revision with the issued version clause by clause. The task analyst rewrites the affected steps, the criticality rater re-rates the task, the question writer redrafts the questions that test the old steps, the lesson planner updates the lesson and slides, and the impact mapper lists who must retrain and by when. In the demo, that took 11 minutes for three changed clauses.
The criticality rater proposes difficulty, importance and frequency, each scored 0–4, using incident history for importance and switching-schedule counts for frequency. A score of 10 or more, or importance 4, makes a task critical. The rater only proposes; three SMEs confirm a critical rating.
It keeps the task list, the documented method it was built with and its annual review date, and sets the practical re-assessment for everyone on a changed task six months from the revision, as NERC PER-005-2 R3.1 describes. In the demo, all 64 people on the changed tasks are booked on the rig by 9 Apr 2027. Your compliance team decides how you use it as evidence.
For each changed or new step, the question writer drafts knowledge questions, scenarios on named equipment and observed practical checks. Distractors come from the old procedure and from real errors in incidents, and every right answer cites its clause. An SME accepts, edits or returns each one with a reason.
The impact mapper reads the authorisation register for everyone who holds an affected task, and the next 7 days of switching schedules. Anyone switching on a changed critical task gets a briefing before their next switch; everyone else re-assesses within six months. The training manager approves the plan.
Yes. The agents read the source systems read-only and only draft and propose. An SME signs every task and question, the instructional designer signs each task analysis and publishes, and the training manager approves who retrains. Agents never edit a procedure, remove a task or suspend an authorisation.
The Agentic Solution Engine builds and deploys it from your requirements and documents: your procedures, task list, rating method, question bank and lesson plans. It goes live once every quality gate has passed. We will walk you through it on one of your own recent revisions first.
We'll run Job Task Analysis on one of your own recent procedure revisions and the task list behind it.