On this page
Maintenance teams can use AI to turn rough fault notes into clean log entries, turn long equipment manuals into short checklists, write clear shift handover notes and build spare-parts lists in Excel. These are writing and sorting jobs, and AI does them fast. It never replaces the manufacturer's instructions, your lockout and isolation procedure, or the judgement of the technician standing at the machine.
TL;DR:
- Use AI on the paperwork around maintenance: logs, checklists, handovers, parts lists and monthly reports.
- Always feed it the real source: the manual, your records, your technician's notes. Never ask it for torque, pressure or clearance values from memory.
- Isolation, lockout and permits to work follow your written procedure. AI is never part of that step.
- In Kenya, some equipment needs statutory examination by an approved person. AI can remind you of dates you enter, not replace the examination.
- Keep machine settings, recipes and staff names out of public AI tools.
Who this is for
This guide is for maintenance technicians, supervisors and engineers in Kenya who look after machines and buildings: production lines, generators, pumps, boilers, lifts, cold rooms, kitchen equipment, hospital plant and estate services. If you look after vehicles, read how fleet managers can use AI as well. For the factory-wide picture, see how manufacturers in Kenya use AI on the factory floor.
Each prompt uses the four Cs we teach at FuKazee: Clarity (what you want), Context (the equipment and your site), Constraints (what to leave out) and Customisation (your format and your words).
Six tasks where AI helps today
1. Fault log entries from rough notes
Technicians often write in a hurry, or send a voice note at the end of a long job.
Below are rough notes from a technician about a fault on [Line 2 filling machine], with names replaced by codes. Turn them into a fault log entry with these fields: date and time reported, equipment, symptom, what was checked, what was found, action taken, parts used (with part numbers if given), time the machine was down, status (fixed, temporary fix, waiting for parts), and follow-up needed. Write NOT STATED where a fact is missing. Do not guess the cause. Notes: [paste]
Check: the technician who did the job reads the entry. Times, part numbers and the status must be exact, because next month's decisions rest on these logs.
2. Checklists from the manufacturer's manual
Manuals are long. The person doing the daily check needs one page.
Below is the maintenance section of the manufacturer's manual for our [air compressor]. Turn it into three checklists: daily, weekly and monthly. Each line: what to check, how, what "OK" looks like, and the manual page or section number. Use only the steps in this text, in the same order. Do not add steps, values or tools that are not in the manual. If something is unclear, write [CHECK MANUAL]. Text: [paste]
Check: a senior technician compares the checklist line by line with the manual. Keep the manual itself on hand, because the checklist is a reminder, not a replacement. Where the manual and AI disagree, the manual wins.
3. Shift handover notes
Turn these notes from the night shift maintenance team into a handover note for the morning team, under one page. Sections: equipment still down and why; temporary fixes in place that need a proper repair; parts ordered and expected; isolations still in place (equipment and who holds the lock); safety concerns; first jobs for the morning shift. Do not leave out any open item. Notes: [paste]
Check: the outgoing and incoming supervisors go through it together, in person, at the equipment where needed. Any isolation still in place is handed over under your lockout procedure, never by a note alone.
4. Spare-parts lists in Excel
I am building a spare-parts list in Excel for our maintenance store. Columns: part number, description, equipment it fits, location in store, quantity in stock, minimum level, reorder quantity, supplier (leave blank), lead time in days. Give me a formula that writes "ORDER" when quantity is below the minimum. Then show me how to filter for every part marked ORDER and sort by lead time, longest first.
Then, once your list is filled in from your own records:
From the parts list below, which critical parts have only one in stock and a lead time over [14] days? Present a short table. Do not change any figure.
Check: part numbers come from the manual or the supplier, never from AI. Count the store before you trust the list. Our Excel AI training page and our guide to Copilot in Excel cover formulas and checks.
5. A preventive maintenance calendar
Below are the service intervals from the manuals for our [six] main machines, the date of each last service, and the dates of any statutory examinations from our records. Build an Excel calendar for the next twelve months showing what is due each week. Add a column that turns red when something is due within 14 days. Use only the intervals and dates I give you.
Check: every interval comes from the manual or your engineer. AI only arranges what you give it.
6. A monthly maintenance report
Using the fault log below for [October], show me which five machines had the most downtime in hours, the most common symptoms, and how many jobs are still waiting for parts. Present a table and four plain sentences for the operations manager. Do not suggest causes that are not in the log.
Check: total downtime in your summary matches the log. A machine with many short stops may matter more than one long stop. Talk to the operators before you report a pattern.
What AI must never do in maintenance
- Replace the manufacturer's instructions. Values such as torque, pressure, clearances, fluid types and intervals come from the manual or the manufacturer. AI can sound sure and be wrong.
- Replace isolation, lockout or permits to work. These follow your written procedure and are done by authorised people at the equipment. Never let a chatbot tell someone a machine is safe to work on.
- Diagnose dangerous equipment alone. Electrical faults, pressure systems, lifts and gas need qualified people. Use AI to organise notes, not to decide.
- Sign off work. The responsible technician or engineer signs.
- Hold your secrets. Machine settings, recipes and supplier prices are business secrets. Staff names in logs are personal data under the Data Protection Act, 2019. Use codes in public tools, or your company's approved tool.
Kenyan rules maintenance teams should know
The Occupational Safety and Health Act, 2007 covers machinery and plant. As of November 2026, from the sections we read:
- Section 60: all fencing or other safeguards must be "constantly maintained and kept in position while the parts required to be fenced or safeguarded are in motion or in use".
- Section 65(2): parts and working gear of every lifting machine must be thoroughly examined at least once in every twelve months, or after modifications or extensive repairs, by a person approved by the Director.
- Section 67(8): every steam boiler and its fittings must be thoroughly examined by an approved person at least once in every twelve months, and after modifications or extensive repairs. The examiner may set a later date, of no more than fifteen months.
- Section 67(5): no one may enter a steam boiler that is one of a range unless it is disconnected from the others, or the valves are "closed and securely locked". This is one example of lock-off written into the law itself.
The Directorate of Occupational Safety and Health Services (DOSHS) lists examination and testing of steam boilers, steam and air receivers, lifts, gas cylinders and cranes among its roles. Check the Act and ask DOSHS about your own equipment. This is general information, not legal or engineering advice.
How FuKazee trains maintenance and technical teams
FuKazee's training for engineering teams and manufacturing companies covers this kind of work. As those pages explain, teams practise first drafts of maintenance procedures and handover documents that the responsible engineer then reviews and signs off, and Excel work to clean machine logs and find which line loses the most hours, with Copilot if you have it and free tools if you do not. Exercises use realistic invented or masked material, so nobody needs to share real drawings or client files. Training runs at your plant by arrangement, or live online.
Questions people ask
Can AI read an equipment manual for me?
It can summarise a manual you paste or upload and turn it into checklists. Keep the original manual and check every line, because summaries can drop a warning or change a number.
Can AI explain a fault code?
Look the code up in the manufacturer's manual or ask the supplier. AI may give a general meaning that does not match your model. Use it to organise the manual's own fault table into a quick reference.
Can technicians use AI on a phone in the workshop?
Yes, for notes and logs, when they are away from moving machinery. Writing up a job after it is finished is safer and gives better notes.
Can AI predict when a machine will break down?
Specialist software can analyse sensor data for this, if you have it. A general chatbot can help you spot patterns in a fault log you give it, which is a good start for most teams.
Sources
All sources last checked 15 November 2026.
- Occupational Safety and Health Act, 2007 (Cap. 236A), Kenya Law official revised edition: sections 60, 65 and 67. The Kenya Law page did not load for us, so we checked the wording in two copies: the Act as passed in 2007 (COTU-K copy) and the revised text on SheriaPlex (section 60, section 65, section 67). Both say the same.
- DOSHS, Our History: DOSHS roles, including examination and testing of boilers, receivers, lifts, cylinders and cranes.
- The Data Protection Act, No. 24 of 2019, ODPC copy: section 2 (personal data).
- FuKazee facts: AI training for engineering teams and AI training for manufacturing companies.
Where FuKazee fits in
FuKazee is a Nairobi training company that has trained more than 3,000 people since 2022. We teach maintenance, engineering and production teams to use AI on logs, checklists and reports, with the free version of each tool first and verification taught as hard as prompting. Tell us what equipment your team looks after and where the paperwork piles up. Email lead@fukazee.com
Want guided support instead?
Not sure which programme fits? Tell us who you are and we will point you to the right path: corporate AI training, the ScholarShift Accelerator for postgraduates, or KCSE study plans and tuition.





