What goes wrong between the warp beam and the greige store
The shed writes most of it down: stops in the loom log, recipe and add-on on the beam card, points on the inspection report. The hard part is putting those records side by side when a loom lags or a roll fails.
Stop reasons written from memory at the end of the shift
Looms stop for warp breaks, weft breaks and mechanical or electrical faults, and any stop can leave a defect in the cloth. When the reason goes on a paper loom log at shift end, it is written from memory. The shift in-charge builds the monthly stop Pareto from that log, so it is only as good as that memory. The weaving manager cannot tell whether a lagging loom has a warp, weft or machine problem.
1.2 stops an hour
Warp-break stops on one loom with 6,000 ends at 210 rpm and 0.2 breaks per 1,000 ends per 10,000 picks: a 1983 worked example from the Technical University of Budapest, at shuttle-era speeds.
Source: Periodica Polytechnica Mechanical Engineering (Technical University of Budapest), Reduction of Warp Breaks on Weaving Looms by Modification of the Mode of Beat-up (opens in a new tab), 1983.
A beam sized off target, discovered only on the loom
Size add-on drifts out of its band: too little and the warp keeps breaking, too much and the beam turns sticky in the shed. Once a beam is drawn in, its add-on cannot change. The playbook’s rule: an over-sized beam can often run with raised RH and lower speed; an under-sized beam eats the day’s efficiency in warp breaks and should be rejected for a buyer-critical lot.
9.94% add-on
In a 2022 Bangladeshi trial on 30 Ne cotton warp, warp breakage was lowest at 9.94% add-on and higher at both 8.93% and 11.23%. That is one yarn’s optimum, not a target for yours.
Source: Journal of Engineering Science (KUET), via Bangladesh Journals Online, Analysis of the Effect of Sizing Add-on% and Sizing Process Parameters on the Average Warp Breakage Rate for 30Ne 100% Cotton Warp Yarn (opens in a new tab), 2022.
Defects graded at the 4-point table, traced to a loom weeks later
Inspectors score each roll and send the mendable defects to mending. Tracing a defect back to its loom, beam, shift and weft lot is done by hand, if at all, so the same fault returns. A weft bar from mixed weft lots may only show after dyeing, when the fabric has left the shed. The inspection in-charge carries the rework, and the owner sells the B-grade rolls cheap.
45–65%
How far fabric with defects tends to fall in selling price: a figure from Selvi and Nasira (2017), repeated in a 2021 peer-reviewed fabric-inspection paper. Not tied to one country or mill.
Source: Sensors (MDPI), Deep Convolutional Neural Network Optimization for Defect Detection in Fabric Inspection (opens in a new tab), 2021.
EPI and PPI checked by hand, and PPI drift that fails GSM later
EPI and PPI are checked with a pick glass, often on the first metres of a beam run only. A take-up or let-off fault can change the picks per inch along the roll. The fabric looks fine at the inspection table, then fails GSM at finishing or a buyer reports it off-spec. To answer, the shift in-charge finds the old readings and compares them with the approved construction by hand.
Five jobs the weaving agent does when someone in the shed asks
It reads what you connect or upload: loom logs, sizing records, 4-point sheets, and Odoo production orders and stock transfers. A new planned start or a message to the sizing master waits for an admin to approve or decline it, and runs only if approved.
A named loom’s efficiency and stop reasons, from the loom log rows
Answers with sources
Upload the loom log or shift report as CSV, or as Excel with the data on the first sheet. Name a loom and shift, and the agent averages its efficiency from the actual rows, or counts its stops for one reason, such as warp break. Ask which looms lagged, and it reads the shift’s rows against the playbook’s target for each type (85% air-jet, 80% rapier, 75% projectile), citing every row it names.
Ask“What was loom 14’s average efficiency on B shift yesterday, and which stops does the log record for it?”
The sizing recipe behind a beam, set against its sizing record
Reads your documents
Upload your sizing recipe sheets and the sizing register as typed files; a beam card on paper, or a photo of one, has no text to read. Ask about a beam and the agent quotes the approved recipe for its construction: size %, viscosity, temperature, stretch %. It sets the beam’s register entry beside it and the playbook’s pre-beam-run checklist, and names each item the record does not show, such as an add-on test.
Ask“Beam 2217 is for the 40×40 / 124×64 quality. What is the approved sizing recipe, and what does the sizing register record for it?”
The week’s defect Pareto table and a root-cause note, as files
Creates files
Keep your 4-point results in one Excel or CSV sheet, a row per defect, with the roll, loom and shift on each row. The agent counts the rows for each defect type, one exact count per type, and ranks them largest first in a spreadsheet, a Pareto table rather than a chart. For a repeat defect, it sets the inspection rows beside the loom log and sizing register in a root-cause note. The weaving manager decides the cause.
Ask“From this week’s 4-point sheet, count each defect type and rank them in a spreadsheet, then list the rows that record slack picks.”
EPI, PPI and width checked against the approved construction
Answers with sources
Upload the pick-glass reading sheet. Name a roll, and the agent works out its average, lowest or highest PPI or EPI from the rows and sets it beside the approved construction and the playbook’s ±2% band. For a buyer dispute, it lays out the playbook’s 9-point report from the sheet: three readings along each roll at three points across it, each traced to its row.
Ask“For roll 412-07, what is the lowest PPI on our pick-glass sheet, and is it within 2% of the 124×64 construction?”
Weaving orders in Odoo, and a new start date for a late one
Reads your ERP
If your factory runs Odoo, the agent reads its production orders (order, product, quantity, status) and stock transfers, with the time each was fetched. When a rejected beam holds up a weaving order, it can draft a new planned start for it. A department admin approves or declines it, and the change runs only if approved. Loom figures never come from the ERP; they come from your loom log.
Ask“Beam 2217 is rejected and the replacement is ready Thursday. Draft Thursday as the new planned start for production order MO/00418.”
A shift, a beam and a buyer: three moments in the shed
Three looms on one construction keep lagging
Shift in-charge
Looms 14, 15 and 22 run the same 124×64 quality and have been below their efficiency target since Monday.
Asks
“Looms 14, 15 and 22 have been below target since Monday. Compare their stop reasons and the beams they are on.”
The agent
Reads the loom log rows for the three looms since Monday and lists each loom’s stops by reason, citing the rows; from the uploaded log it counts one loom’s warp-break stops exactly when asked. Reads the sizing register rows for their beams: recipe, add-on % and test date. Quotes the playbook: when three looms on one construction show the same stop pattern, the issue is upstream, in yarn or sizing.
Result
The shift in-charge takes the three looms’ recorded stops and the sizing records behind them to the sizing master, instead of a guess.
A beam fails its add-on test after drawing-in
Weaving manager
The add-on test on beam 2217 comes back under target after the beam is drawn in, and the beam is for a buyer-critical lot.
Asks
“Beam 2217 tested under its add-on target after drawing-in. What does the playbook say? Then draft a WhatsApp to the sizing master about the next beam on this recipe.”
The agent
Reads the beam’s sizing register entry, the add-on test and the recipe sheet. Quotes the playbook: a drawn-in beam’s add-on cannot change; an under-sized beam on a buyer-critical lot should be rejected; the correction goes on the next beam, re-tested before drawing-in. Lists the next beams on that recipe from your sizing register, and drafts a WhatsApp Business message to the sizing master naming them.
Approval
The message waits for the weaving department admin, who approves or declines it. It goes to the sizing master only if approved, and the audit trail records the decision.
Result
The manager makes the run-or-reject call with the playbook’s rule and the beam’s records in hand. The written correction for the next beam is sent to the sizing master only if the admin approves the message.
A buyer says the PPI on a lot is off-spec
Inspection in-charge
The buyer’s report on lot 412 says PPI is below the construction, and it is uploaded to the weaving department.
Asks
“Our checker took nine readings on each retained roll of lot 412. Make the 9-point report and compare it with the 124×64 construction.”
The agent
Reads the pick-glass reading sheet for the lot and lays out each retained roll’s nine readings in a spreadsheet, beside the construction and the playbook’s ±2% band, citing the row behind every reading. For a roll you name, it works out the average, lowest or highest PPI from the rows. Adds the playbook’s note: if all nine points are in spec, the buyer’s single reading may be the wrong one.
Result
A 9-point report for every roll, each reading traced to the sheet, to set beside the buyer’s report before anyone discusses a claim.
The cost of a stopped loom and a downgraded roll, in published studies
Published figures, cited at the source: a 1983 university calculation, a price figure from fabric-inspection research and one Bangladeshi mill’s records. They are not Mavick results. Record your shed’s KPIs below before a pilot, so it is judged on your looms.
Fewer looms standing on a warp break
About 5%
Machine efficiency lost when warp breaks rise by 0.1 per 1,000 ends per 10,000 picks, in a 1983 Technical University of Budapest calculation for shuttle-era looms. It shows why one more break matters; it is not a benchmark.
How the agent helps
The agent does not touch the looms. When you ask, it averages the warp break rate your loom log records for each loom or beam you name, a few at a time, and sets it beside the playbook’s check order for a lagging loom. Ask beam by beam and it reads each beam’s recipe from the sizing register, so the sizing master can see whether breaks run higher on one recipe.
Track: Warp breaks per 100,000 picks and loom efficiency %, by loom and shift
Source: Periodica Polytechnica Mechanical Engineering (Technical University of Budapest), Reduction of Warp Breaks on Weaving Looms by Modification of the Mode of Beat-up (opens in a new tab), 1983.
Fewer rolls downgraded at the inspection table
45–65%
How far a defective fabric’s selling price tends to fall, an estimate from earlier research that a 2021 peer-reviewed fabric-inspection paper repeats. General, not one mill’s figure.
How the agent helps
When your 4-point sheet records the loom for each roll, the agent counts a defect type for a loom or a shift from its rows. For a repeat defect, it reads that loom’s log rows, sizing record and weft lot record beside the inspection rows. It does not grade cloth. Once the link back to the loom is on the record, a repeat defect can be fixed at the loom, not mended again.
Track: First-pass inspection: rolls accepted as-is
Source: Sensors (MDPI), Deep Convolutional Neural Network Optimization for Defect Detection in Fabric Inspection (opens in a new tab), 2021.
Greige against plan, without reading the report row by row
About 270 rows a day
One day’s production report at a Bangladeshi mill with 172 rapier looms, 2013: about 270 rows and 22 columns, kept as a record, with missing values and typing mistakes, the authors say.
How the agent helps
Upload the report as CSV, or as Excel with the data on the first sheet, one header row and no blank “same as above” cells. The agent sums greige production for a quality you name, or counts cut pieces for a loom, from the actual rows, and shows which rows it used. A table holds up to 5,000 rows, so upload a week or two at a time.
Track: Greige production against plan, by quality and week
Source: Data in Brief (Elsevier), via PubMed Central, Textile weaving dataset for machine learning to predict rejection and production of a weaving factory (opens in a new tab), 2023.
What to baseline before a pilot
Measure each one before the pilot starts, then again at its end.
| KPI to baseline | Why it matters |
|---|---|
| Loom efficiency % by loom and shift | The playbook sets 85% or more on air-jet, 80% on rapier and 75% on projectile. Read loom by loom, it shows the looms a shed average hides. |
| Sizing add-on % against target, by beam | The playbook allows ±1% of target. A beam outside it either breaks too often or runs sticky, and its add-on cannot be changed once it is drawn in. |
| First-pass inspection: rolls accepted as-is | The playbook targets 90% or more. Below 80%, it reads the cause as instability upstream, in yarn or sizing, not at the inspection table. |
| Rolls with EPI and PPI within ±2% of the construction | The playbook’s target is every roll. PPI drift is what lets a fabric that looks right fail GSM at finishing. |
| Defect points per 100 sq yd (or 100 sq m, as each buyer states it), by week | The threshold is the buyer’s, so baseline it against each buyer’s limit. The playbook reviews the defect Pareto every week, top three causes first. |
What the agent works from
Records it reads
- Beam card
- Loom log
- 4-point inspection report
- Mending register
- Sizing recipe sheet
- Sizing register
- EPI/PPI reading sheet
- Daily production report
In your ERP
Reads Odoo production orders (order, product, quantity, status) and stock transfers (partner, status, scheduled date, source document). Drafts a production order’s planned start or assignee, or a transfer’s note or source document, for an admin to approve or decline. Loom, beam and inspection records come from your uploaded logs 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 loom-log or beam-status sheet in Google Sheets when you ask. Drafts a new row for that sheet, the shift handover for the shed’s Slack channel, or a WhatsApp Business message to the sizing master about the next beam. Each waits for an admin to approve or decline it, and none runs until approved.
Tools named today: Gmail, Outlook, Google Sheets, Slack and WhatsApp Business.
Never made up
- Loom speeds, efficiency percentages or stop counts
- Sizing recipe percentages and warp particulars
- EPI, PPI, width or construction values
- 4-point inspection points and roll grades
If it is not in your connected data, the agent says so.
Belongs to another department
- Yarn quality parameters and count testingSpinning
- Greige processing, dyeing and finishing recipesDyeing & Processing
- Loom repair scheduling and spare availabilityMaintenance & Engineering
- Order allocation and ship datesPlanning & PPC
What the Weaving playbook covers
SOP: 4-point inspection and mending
Every roll is inspected on the 4-point system: each defect scores 1 to 4 points by length, with no more than 4 points per linear metre. The roll is accepted or rejected on points per 100 sq yd against the buyer’s threshold, commonly 40 or fewer for first quality. Mendable defects go to mending, and mended rolls are checked in full before re-roll.
Decision rule: a loom below its efficiency target
Check in this order: warp break rate (size add-on, yarn quality, RH), weft break rate (cone winding, weft tension), mechanical stops (beat-up timing, shedding sequence), then the operator. The playbook says most drops resolve at the first or second step; if they do not, escalate to maintenance.
KPI: warp and weft breaks per 100,000 picks
Targets are set by construction; for warp breaks, typically under 3. The playbook reads a rising warp break rate as a sizing or warp yarn problem, and a rising weft break rate as weft cone quality or weft tension.
Heuristic: the quiet rapier
Air-jet looms are loud about their problems; rapier looms are silent. The playbook warns that the rapier with a slowly rising break rate is the one that will surprise you. Its KPIs track efficiency per loom and per shift, not only for the whole shed.
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.
Loom efficiency and stoppages
Show loom-wise efficiency for the last shift with the stoppage reasons logged, and tell me which looms are furthest below target.
Approved sizing recipe for a beam
What is the approved sizing recipe and warp particulars for this beam, and what was actually applied?
4-point inspection on a roll
What did the 4-point inspection record for this roll? List the defects by type and point count, and give me the grade.
EPI and PPI against the construction
Do the EPI, PPI and width on the quality currently running match the approved construction? Show both sets of figures.
Beams loaded and pending
Which beams are currently loaded, which are pending, and what does that mean for the orders in the plan?
Root-cause a recurring defect
This defect keeps recurring. Pull the loom log and inspection reports for the affected looms and tell me what the records point to.
Greige production against plan
Show greige production against plan for the last week by quality, and flag any quality running short.
Mending load pending
How much fabric is pending in mending, how long has it been waiting, and which orders does it belong to?
Warp breakage trend
Show the warp breakage rate over the recent beams for this quality and tell me whether it is trending with any sizing recipe change.
Shift handover for the shed
Prepare a shift handover for the weaving shed from the loom logs: what ran, stoppages of note, beams changed, and what the next shift must watch.
Weaving: questions before a demo
Our loom log is still on paper. Can the agent work from it?
Only once it is typed in. The agent reads files that contain text, so a photo or a scan of a loom log sheet gives it nothing to read. When the shift’s stops and production go into a Google Sheet, or an Excel or CSV file you upload, it answers from those rows, and from an uploaded Excel or CSV table it works out efficiency and stop counts. Sizing recipes, the sizing register and SOPs that already exist as typed files (not scans) can be uploaded as they are, so you can start there.
Can it read loom data straight from our looms or our loom-monitoring system?
No. Mavick does not connect to looms, sensors or loom-monitoring systems. If your monitoring system exports a stop report to Excel or CSV, upload it and the agent counts the stops for a reason you name from its rows. No ERP gives the weaving agent loom or beam records, Datatex NOW included: from the factory’s ERP it reads Odoo production orders and stock transfers only. Loom speeds, efficiency and stops reach it through your exported reports or a Google Sheet.
Can it tell us which beam to load on which loom next?
No. It does not plan or allocate looms. Ask which beams are loaded and which are pending, and it lists them from your beam-status sheet or sizing register. If you run Odoo, it also reads the production orders for those qualities: order, product, quantity and status. Order allocation and ship dates stay with Planning & PPC, and the weaving agent says so when asked.
Can the agent change a sizing recipe or a loom setting?
No. It does not write recipes, loom settings or quality records to any ERP, and it does not connect to the sizing machine or the looms. The ERP changes it can draft are narrow and only in Odoo: a production order’s planned start or who it is assigned to, or a note or source document on a stock transfer. Each waits for an admin to approve or decline it and runs only if approved. For a recipe correction, it drafts the message to your sizing master instead, sent only if approved.
Can it find defects in the fabric or grade our rolls?
No. It does not see the cloth, and it does not work with cameras or inspection machines. It reads the 4-point results your inspectors record. From results uploaded as an Excel or CSV sheet, it totals a roll’s points or counts a defect type, for a loom or shift when the sheet records them. A PDF report it reads and quotes. It sets each roll’s score beside the buyer’s threshold, once that threshold is in a document you uploaded. The grade stays your inspector’s.
Will the weaving agent see Spinning’s yarn test reports?
No. Spinning’s USTER and HVI files stay in Spinning; the weaving agent works from what is connected to Weaving, such as your loom log, sizing register and 4-point sheets. Asked about yarn quality parameters or count testing, the weaving agent says Spinning owns them. If a yarn test certificate comes in with a weft lot and you upload it to Weaving, or share it across the factory, the weaving agent reads that copy.
Bring Weaving. 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)