Breakdowns, slipped PM, utility bills and missing spares
The records usually exist: a PM schedule, a breakdown register, the meter readings, a spares list. Nobody adds them up until a machine stops, and then the cost lands in lost production hours, night repairs and the utility bill.
PM jobs that slip until the machine fails
The PM schedule sits on a board or in an Excel file. When production is busy, jobs get pushed back, and nobody can say which machines are overdue or by how long. The playbook reads PM completion under 90% for a quarter as a programme on paper only. The maintenance head has to explain the gap at the review, and the night-shift fitters get the breakdowns.
15–30%
Share of steam traps that can be failing after 3 to 5 years without maintenance, against under 5% with a scheduled programme. US DOE figures for steam systems in general, quoted in a 2010 guidebook for the textile industry.
Source: Lawrence Berkeley National Laboratory (ENERGY STAR), Energy-Efficiency Improvement Opportunities for the Textile Industry (opens in a new tab), 2010.
Utility cost per kg that nobody can explain
Compressors, boilers, the dye house and the ETP are the biggest energy users, yet their meter readings rarely sit in one place. Energy per kg creeps up through air leaks, failed steam traps and boiler scale, and the bill arrives with no cause attached. When the grid or the gas fails, the diesel generator runs at part load. The utilities in-charge and finance both pay for it.
Up to 5%
Fuel that scale on the tubes can waste in a fire-tube boiler; in a water-tube boiler, 1 mm of scale can raise fuel use by 2%. US and Canadian test figures, quoted in a 2010 guidebook for the textile industry.
Source: Lawrence Berkeley National Laboratory (ENERGY STAR), Energy-Efficiency Improvement Opportunities for the Textile Industry (opens in a new tab), 2010.
A breakdown register that nobody adds up
Each stop goes into the register: machine, fault, start and finish, parts used. Nobody works out MTTR or MTBF from it, so a repeat fault goes unseen and the patch-or-overhaul call is made from memory. The fix sits with one senior fitter or in an OEM manual in a cupboard, so the night shift diagnoses the same fault again while production loses the hours.
The spare that is not there when the machine is down
A bearing, a belt or an electronic card fails, and the part was never on the critical-spares list. Imported OEM spares can take weeks to arrive. The machine waits, the store and purchase get an urgent request, and PPC re-plans the orders on it. A bearing job of a few hours turns into a stoppage of weeks.
Overdue PM, MTTR and spares, worked from your logs
It answers from the PM schedule, breakdown log, spares sheet, energy report and OEM manuals you upload or connect, and from Odoo stock transfers, citing what it used. A message to a fitter or a service agent waits for an admin to approve or decline it.
Overdue PM, listed and drafted for the fitter who owns it
Acts in your tools
Ask which PM jobs are past due and the agent reads the next-due dates in your PM schedule, from an uploaded Excel file or a connected Google Sheet. It lists the jobs by section and machine with the last-done date, citing the rows. It can draft that list as a WhatsApp Business or Slack message to the section fitter. A department admin approves or declines the message, and it is sent only if approved.
Ask“Which PM jobs in the spinning section are past due? List them by machine, then draft the list for the fitters on WhatsApp.”
MTTR, MTBF and repeat faults, worked out from the breakdown log
Answers with sources
Upload the breakdown log as an Excel or CSV file. For the machines and faults you name, the agent counts breakdowns, adds up downtime hours and averages the logged repair time for MTTR, citing the rows. Repair time needs its own column; start and finish times alone are not enough. MTBF needs run hours: if they are not in your data, it says so. A repeat fault counts only when it is written the same way each time.
Ask“From the October breakdown log, what was MTTR for stenter 2, and how many of its stops were burner faults?”
The OEM manual and your SOPs, quoted on any shift
Reads your documents
Upload OEM manuals, SOPs and PM checklists as text PDFs or Word files. Ask for a service interval, an oil grade or a belt-tension check, and the agent quotes it and names the document it came from. If the manual does not give it, the agent says so: it never makes up a manufacturer setting or tolerance. It cannot read wiring diagrams or exploded views, only text and tables.
Ask“What does the OEM manual give as the oil and filter change interval for compressor 2?”
The spare on your stock sheet, its receipt in Odoo
Reads your ERP
For a pending job, the agent reads the spare’s on-hand quantity and min level from your spares sheet, uploaded as Excel or CSV or kept in a Google Sheet, citing the row. Given a PO number or supplier, it reads the matching Odoo stock transfer’s status and scheduled date, with the fetch time. It can draft a note on that transfer naming the machine and job, for an admin to approve or decline.
Ask“Is the 6205 bearing for loom 14 above its min level on the spares sheet, and when is the Odoo receipt for PO 4471 scheduled?”
Energy per kg, built as a sheet from your meter readings
Creates files
Upload an energy report covering both months, with a month column, as an Excel or CSV file. The agent totals kWh, steam and gas by utility for this month and last, from the rows, and divides by production kg when that figure is in your data. It builds the comparison as a spreadsheet and sets out the playbook’s three checks for a jump: production volume, load mix, then plant losses such as leaks or scale.
Ask“Build October’s utility sheet: kWh, steam and gas by utility against September, and per kg where we have production.”
A month-end review, a repeat trip and a steam bill
The monthly maintenance review
Maintenance head
The morning of the monthly review with the plant head, with October’s PM sheet and breakdown log uploaded.
Asks
“From the October PM sheet and breakdown log, give PM completion and breakdown hours by section, and list the breakdowns in the section that lost most.”
The agent
Counts PM jobs done against jobs scheduled for each section from the PM sheet, and adds up each section’s breakdown hours from the log, citing the rows. For the section with the most hours, it lists the breakdowns by machine and fault, so a repeat fault stands out. Quotes the playbook: completion under 90% for a quarter goes to the COO, and a machine failing repeatedly within a month needs an overhaul, not another patch. Builds the review as a spreadsheet.
Result
One sheet for the review, each section’s figures calculated from the rows, and the overhaul candidates named from the record.
A compressor trips for the third time
Utilities engineer
Compressor 2 trips on high discharge temperature for the third time this month, and the plant is running on the standby unit.
Asks
“Compressor 2 tripped again. List its breakdowns this month, check stock of the oil filter and thermostatic valve kit, and draft an email to the service agent asking for an overhaul visit.”
The agent
Filters the uploaded breakdown log on its machine and month columns: three trips for compressor 2 this month, each with its fault, repair time and parts used, rows cited. Reads the on-hand quantity of both parts from the uploaded spares sheet, citing the rows. Quotes the causes of high discharge temperature from the uploaded OEM manual, naming it. Prepares the email to go out through Outlook, with the dates and faults as logged.
Approval
The maintenance head decides as department admin: approve or decline. The email is sent only if approved, and the audit trail keeps who decided.
Result
If approved, the email is sent with the three logged trips in it, and the patch-or-overhaul call rests on the record rather than memory.
Steam per kg is up at month-end
Energy manager
The October gas bill is up, and the plant head wants the reason before the monthly review.
Asks
“Steam per kg rose in October. Compare boiler steam, gas and production kg with September and show where it moved.”
The agent
Adds up steam and gas for each month from the uploaded energy report, filtered on its month column, citing the rows. If the report carries production kg, it sums that the same way and divides; if not, it says so. Works through the playbook’s three checks in order: volume, load mix, plant losses. It answers load mix only if the report records it. Quotes the boiler-scale heuristic and builds a steam-trap and leak survey sheet for the next outage.
Result
A month-on-month steam-per-kg sheet with its rows, the checks the records can answer, and a survey sheet for the planned outage.
What a stopped machine, an air leak and a diesel hour cost
Published figures from one apparel plant, US energy guidance and Pakistan’s power market, cited at the source. None is a Mavick result. For your own numbers, a pilot records the KPIs below first and reads them again at the end.
Repairs that follow the written fix
13 to 5 minutes
At one Sri Lankan apparel plant, a thread-slipping breakdown on a flatlock machine took 13 minutes to repair. With a standard way of attending breakdowns, the paper reports 5 minutes “for one machine timing setting”. One line, 2010 data.
How the agent helps
The agent does not repair machines. When you ask, it quotes the written fix from your SOPs and OEM manuals, naming the document, and works out MTTR by machine and fault from the breakdown log’s rows. Once the senior fitter’s fix is written down and uploaded, the night shift can find it too.
Track: MTTR by section and type of fault
Source: Prabandhan: Indian Journal of Management, Developing A Total Preventive Maintenance (TPM) Plan For Flatlock Sewing Machines: A Case Study In The Apparel Industry In Sri Lanka (opens in a new tab), 2012.
Air, steam and power per kg, from your own readings
20–30%
Share of a compressor’s output that leaks often waste in industrial compressed-air systems. A 2004 US Department of Energy tip sheet, across industries, not a textile figure.
How the agent helps
The agent reads no meters and finds no leaks. From an uploaded Excel or CSV energy report with a week or month column, it totals compressor kWh and steam for each, and use per kg when production kg is in it, showing the rows. The playbook asks for the cause of any weekly change above 5%, and names plant losses such as an air leak or boiler scale as the usual cause of a jump.
Track: Energy and steam per kg, by month and fabric type
Source: U.S. Department of Energy, Minimize Compressed Air Leaks (Compressed Air Tip Sheet #3) (opens in a new tab), 2004.
Diesel hours priced against the grid
PKR 120 per kWh
Cost of diesel generation in Pakistan in 2026, with diesel at PKR 380 a litre. Commercial and domestic grid power above low-use tariffs costs PKR 40 per kWh, and industrial power 10–20% less.
How the agent helps
Upload the generator log as an Excel or CSV file with run hours, diesel litres, diesel cost and kWh. The agent totals each for the month from the rows, divides cost by kWh, and sets the result beside the tariff on your uploaded power bill. The playbook puts diesel at typically 2–4 times grid, to be checked against your own tariff, and says to plan for critical loads only.
Track: Diesel cost per kWh against the grid tariff
Source: Ember, The solarisation of Pakistan’s energy economy (opens in a new tab), 2026.
What to baseline before a pilot
Measure each one before the pilot starts, then again at its end.
| KPI to baseline | Why it matters |
|---|---|
| PM-completion % on schedule | The playbook targets 95% or more. Under 90% for a quarter, it calls the programme decorative and takes it to the maintenance head and the COO for a resourcing review. |
| MTTR by type of fault | Shows whether repairs stay inside the playbook’s targets for each kind of fault: 30 minutes for a shift technician’s fix, 4 hours for a gear, bearing or motor job. |
| MTBF by machine class, quarter on quarter | The playbook wants it rising slowly. Flat MTBF, it says, means PM is checking boxes rather than preventing failures. |
| Breakdown downtime as % of available production hours | The playbook’s target is 3% or less. It is the number PPC and the plant head feel as lost output. |
| Stock-outs on critical spares | The playbook targets zero, and reviews the critical-spares list once a year against the breakdown history. |
| Energy, steam and water per kg, by fabric type | The playbook takes a monthly rise above 10% to the energy manager and the maintenance head, with the root cause due within two weeks. |
What the agent works from
Records it reads
- PM schedule
- Breakdown log
- Spare-parts master
- Energy report
- OEM manual
- Pre-outage checklist
- Generator (DG) log
- Boiler log
In your ERP
Reads Odoo stock transfers, such as a spare’s PO receipt (transfer, partner, status, scheduled date, source document), and production orders (order, product, quantity, status). Drafts a transfer note or source document, or an order’s planned start or assignee, for an admin to approve or decline. Work orders and spare stock come from files or a Google Sheet.
One ERP connection serves the whole factory: SAP Business One, SAP S/4HANA, Odoo, Oracle NetSuite, Microsoft Dynamics 365, WFX Cloud ERP or Datatex NOW. This agent reads only the records named above; we check yours on the call.
In your tools
Reads a PM or meter sheet in Google Sheets, and a service agent’s emails in Gmail or Outlook, when you ask. Drafts an overdue-PM list as a WhatsApp Business or Slack message to a fitter, a change marking a job done in the PM sheet, or an email to a service agent. An admin approves or declines each one; it runs only if approved.
Tools named today: Gmail, Outlook, Google Sheets, Slack and WhatsApp Business.
Never made up
- Machine specifications, tolerances or manufacturer settings
- PM schedules, service intervals or job completion status
- Breakdown durations, MTBF and MTTR figures, or downtime causes
- Energy consumption readings and utility costs
If it is not in your connected data, the agent says so.
Belongs to another department
- Production output impact and re-planningPlanning & PPC
- Spare purchases and vendor termsSourcing & Procurement
- Spare stock positions and issuing sparesStores & Inventory
- Safety incidents and permits to workHealth, Safety & Environment
What the Maintenance & Engineering playbook covers
SOP: breakdown response, by who can fix it
An electrical fault or a slipping belt goes to the shift technician: fix in 30 minutes. Gear, bearing and motor jobs go to the mechanical team, within 4 hours. A major failure needing the OEM may take days. On a critical-path machine, the maintenance head, plant head and PPC are told in the same shift, the OEM is called within 4 hours, and a spare machine is named within 8.
Decision rule: a PM window that clashes with peak production
Agree the window with PPC and move it to a weekend or the night shift. The playbook does not allow skipping PM: a job may be deferred by one week at most. Before a planned outage, the playbook’s checklist asks for PM tickets prepared, spares in hand, a re-plan signed by PPC, safety isolation reviewed, and the outage length committed in writing.
KPI: zero stock-outs on critical spares
Keep a critical-spares list for each machine, from OEM data and experience, with min and max levels. A spare with a lead time over 30 days carries 1.5 times cycle stock. Review the list once a year against the breakdown history, because the part that runs out is the one nobody listed.
Heuristic: boiler scale shows in steam per kg, not as a breakdown
Scale raises steam use slowly, over months, while every machine keeps running. The playbook says to descale with acid at the next planned outage and to check the feed-water treatment. Steam per kg rising while production and load mix hold steady is the sign to look for.
Written as general textile practice and reviewed by four independent AI models. Not yet signed off by a human subject-matter expert. Upload your own SOPs and standards, and every answer shows which source it used.
Ten questions it ships with
The template’s starter questions. Your team can edit them, or ask in its own words.
Preventive maintenance due
Which preventive maintenance jobs are due or overdue, on which machines, and how far past schedule is each?
Breakdown history for a machine
Show the breakdown history for this machine: dates, durations, faults recorded, and whether the same fault recurs.
MTBF and MTTR for a section
What are MTBF and MTTR for this section over the last month, and how do they compare with the previous month?
Spare availability for a job
Is the spare needed for this pending job in stock, on order, or unavailable? Show the stock position and any open PO.
Energy consumption by utility
Show energy consumption by utility for this month against last month, and tell me where consumption moved.
Downtime by cause this month
Break down total machine downtime this month by cause, and rank the causes by hours lost.
Breakdown post-mortem
Prepare a post-mortem for this breakdown from the logs: what failed, how long it took, what was done, and what the records suggest about prevention.
Machines with repeat failures
Which machines have failed more than once with the same fault this quarter? Show the occurrences.
Shift handover for engineering
Prepare a shift handover from the maintenance logs: jobs completed, jobs in progress, machines still down, and what the next shift must pick up.
Utility cost per unit of output
Show utility cost per unit of output for this month from the energy and production records, and compare it with last month.
Maintenance & Engineering: questions before a demo
Our breakdown register is on paper. Can the agent still work out MTTR?
Only once it is typed in. A photo or a scanned page of the register has no text to read. Upload the log as an Excel or CSV file, one row per breakdown, with the repair time in its own column; in Excel, keep it on the first sheet. The agent then averages repair time by machine or section from those rows and lists them; for figures by month, add a month column. A connected Google Sheet is read and cited too, but calculations from the rows run on an uploaded file. Write machine names and faults the same way each time, or repeat faults will not count together.
Can it read our PLCs, energy meters or the compressor controller?
No. Mavick does not connect to machines, controllers, meters or a building management system. It works from readings once they are in a file you upload or a Google Sheet. The factory’s ERP gives Maintenance no machine readings, only Odoo production orders and stock transfers. If a reading is not in your connected data, the agent says so instead of estimating it.
Will it remind the fitters when a PM job falls due?
No. The agent answers when someone asks. It does not keep the PM calendar, raise PM tickets or send reminders between questions. Ask which jobs are past due and it lists them from your PM schedule. It can then draft the list as a WhatsApp Business or Slack message to the section fitter, which an admin approves or declines. The message is sent only if approved.
We keep work orders in SAP PM or a CMMS. Can the agent read or open them?
Not directly. Mavick’s ERP connections do not cover maintenance work orders, in SAP or any other ERP, and there is no CMMS connection. Export the open work orders to Excel or CSV, or keep them in a Google Sheet, and the agent reads them there. It does not raise work orders, requisitions or POs, or change stock. In Odoo it can draft a note on the stock transfer that issues a spare, naming the machine and job; an admin approves or declines it, and it is added only if approved. Only Odoo takes that note: with another ERP, Maintenance drafts no ERP change.
Can it tell us which machine will fail next?
No. It does not predict failures or check machine condition. It works from what your breakdown log records: which machines failed more than once with the same fault, and MTBF where run hours are logged. The playbook reads flat MTBF as PM that only checks boxes, and falling MTBF as the time to cost an overhaul against the patch budget. The agent sets that guidance beside your own figures.
Can it read the OEM manuals for our looms, frames and compressors?
Yes, when the file has text. Upload text PDFs or Word files of up to 10 MB each, to Maintenance or shared across the factory, and split a larger manual into parts. A scanned manual has no text to read, and the agent cannot read wiring diagrams, exploded views or photos. It quotes the passage an interval or setting comes from and names the manual. If the manual does not give it, the agent says so rather than guessing.
Bring Maintenance & Engineering. We’ll build it on the call.
Book a 30-minute demo (opens in a new tab)- support@mavick.si
- Demo calendar
- calendly.com/safock/30min (opens in a new tab)