Skip to content

Knitting: needle lines and GSM drift, traced to machine and shift

The knitting agent answers from your program sheets, GSM cards, roll cards and inspection reports, and traces a needle line to its machine. It shows the files, records and rows it used, and any message or ERP change waits for an admin to approve or decline it.

Works from: Knitting program, GSM card, Roll card, Inspection report

Where a knitting floor loses rolls and machine hours

The evidence sits on the roll card, the GSM card, the shift sheet and the program file. Much of it is kept on paper or in Excel and read one roll at a time, so the pattern behind a fault can stay hidden until the dye house or the buyer finds it.

  • Needle lines and oil stains found after the fabric is knit

    A hooked needle drops a stitch now and then, and the line runs down the roll before inspection sees it. Caught at half-roll, the roll goes to the quality queue: a mendable fault goes to a lower-grade buyer, a continuous one is scrapped. Oil from an over-dosed tray is worse: it may show only after pre-treatment, when the dyeing cost is already spent. The in-charge then traces machine and shift from the roll records.

    27%

    Share of defects on 1×1 rib grey fabric that were stains, at one Bangladesh circular knitting mill over three months of 4-point inspection, in a 2020 study. It traces stains to oil-soaked fluff in the needle tricks.

    Source: Communications in Development and Assembling of Textile Products, Exploring weft knit fabric defects based on their presence and quality impact: A case study (opens in a new tab), 2020.

  • GSM that moves from roll to roll in one run

    The swatch passed, but GSM shifts across the rolls: a cone from another lot, take-up tension that crept, a cam reset after an operator change. In one published study, yarn was also the largest cause of knit fabric defects, and the playbook checks yarn first. On a confirmed order, it says to freeze the lot, segregate it and retest, and to tell the QA head and the knitting manager the same day.

    51%

    Share of fabric defects related to yarn quality on one Bangladesh knitting production floor, in a 2022 Six Sigma study.

    Source: Textile & Leather Review, Implementation of the Six Sigma Methodology for Reducing Fabric Defects on the Knitting Production Floor: A Sustainable Approach for Knitting Industry (opens in a new tab), 2022.

  • Machine hours lost to stops nobody adds up

    Yarn breaks, needle breaks, program changes and handling time each stop a circular machine. The reasons go on the shift sheet or the stop log, machine by machine, and are rarely totalled. At the end of the week output is short of plan, and it is hard to say which machines or which causes cost the most. The playbook sets 80% or more efficiency on circular machines, tracked per machine.

  • The approved program on file, and the setting on the machine

    The knitting program fixes the gauge, machine type, stitch length or cam setting, RPM, yarn count and target GSM. On the floor, the setting changes with each new style, and can drift after an operator change. A new style needs a sample roll before bulk. When two machines on one program show the same fault, the playbook says to look at the program. Why a version was approved often sits with the knitting master.

What the knitting agent works out from your roll and GSM records

It reads the program sheets, GSM cards, roll cards and inspection reports you upload or keep in Google Sheets, and Odoo production orders and transfers. It is not connected to the machines. Anything it sends or writes waits for an admin to approve or decline.

  • GSM across a run, worked out from the GSM card

    Answers with sources

    Upload the GSM card as an Excel or CSV file. For the run and machine you name, the agent works out the average GSM from the actual rows, or the lowest or highest when you ask for it, and sets it against the program’s target. It then reads the roll rows and quotes the readings it finds outside the tolerance you give it, citing each row.

    Ask“What is the average GSM for run R-2207 on machine 9, and which roll readings sit outside ±5% of the program’s target?”

  • The approved program, set beside the last setting record

    Reads your documents

    Upload your knitting programs and machine setting records. Ask for a structure and GSM, and the agent quotes the approved program: gauge, machine type, stitch length or cam setting, RPM, yarn count and target GSM. Ask what is running on machine 14, and it sets the last recorded setting beside the program and lists every value that differs or was never recorded. It reads the record, not the machine.

    Ask“What is the approved program for 26/1 single jersey at 160 GSM, and does the last setting record for machine 14 match it?”

  • The greige lot inspection summary, built as a spreadsheet

    Creates files

    From the per-roll inspection reports for a greige lot, the agent builds the summary as a real spreadsheet: defects grouped by type, the rolls accepted, rejected or diverted, and the machine each roll came from when the report records it. Ask how many rolls carry one defect, and it counts them from the rows of an uploaded Excel or CSV. Asked to add each roll’s grade, it edits that part only, with every revision kept.

    Ask“Summarise greige lot G-0418 as a sheet: the defects by type, the rolls rejected or diverted, and the machine for each roll.”

  • One yarn lot per machine, noted on the yarn transfer

    Reads your ERP

    The playbook’s pre-run checklist asks for cones from a single lot. If your ERP is Odoo, the agent drafts a note on the stock transfer that issues the run’s yarn, such as “issue lot 7731 only to machine 14”. A department admin approves or declines it, and it runs only if approved. Without Odoo, it adds the same line as a row in the yarn issue sheet you keep in Google Sheets, which waits for the same approval.

    Ask“Machine 14 starts program P-118 tomorrow. Draft a note on its yarn transfer: issue lot 7731 only, no mixing.”

  • The shift handover, sent to the next supervisor if approved

    Acts in your tools

    Before shift change, ask for the handover. The agent writes it from the shift sheet, the GSM card and the needle-change log: output against target by machine, rolls on hold for GSM, needles changed, and machines idle with the reason logged. It drafts a WhatsApp Business or Slack message to the incoming supervisor, which waits for an admin to approve or decline it and is sent only if approved.

    Ask“Write the night shift handover from today’s shift sheet and needle-change log, and send it to the night supervisor on WhatsApp.”

An oil complaint, a GSM hold and a short shift

  • The dye house reports oil marks on a lot

    Knitting in-charge

    The dye house sends a complaint on part of a lot: oil marks showed up after scouring and bleaching.

    1. Asks

      “Rolls 3318 to 3326 showed oil marks after scouring and bleaching. Which machines and shifts knit them, and what did our inspection record?”

    2. The agent

      Reads the roll cards and inspection reports in the knitting department’s Google Sheet, matches each roll to its machine, shift and knitting date, and lists what knitting inspection recorded for it. It shows whether the rolls share one machine. It quotes the playbook’s edge case: oil that shows only after pre-treatment usually traces to one machine with an over-dosed oil tray.

    3. Result

      The in-charge knows which machine’s oil tray to check first, with each roll record cited, and answers the dye house from the records.

  • GSM drifting low on a confirmed order

    Shift in-charge

    Mid-shift, the GSM readings on machine 9 run low on rolls for a confirmed buyer order.

    1. Asks

      “Machine 9 is reading low on order 5512. List the rolls outside ±5% of target and draft the hold message to the QA head.”

    2. The agent

      Reads the knitting department’s GSM sheet in Google Sheets for machine 9, quotes each roll’s head, middle and tail readings with the rows cited, and points out the rolls whose readings it finds outside ±5% of the program’s target. It sets the yarn issue sheet beside the roll times to show whether a cone or lot change came first, the playbook’s first check. It drafts a WhatsApp Business message to the QA head.

    3. Approval

      The message waits for the knitting manager, as department admin, who approves or declines it. It is sent to the QA head only if approved, and the audit trail records who decided.

    4. Result

      If the manager approves it, the WhatsApp message to the QA head is sent with the held rolls and their readings, the same-day word the playbook’s escalation asks for. If declined, nothing is sent.

  • The night shift came in short of plan

    Knitting manager

    The morning production meeting, after a night shift that fell short of its target.

    1. Asks

      “Show each machine’s output for last night against target, and the stop reasons logged for the three lowest machines.”

    2. The agent

      Reads the uploaded shift sheet and quotes each machine’s kg against the target kg the sheet records, citing the rows. For any machine you name, it totals the night’s kg from the rows. For the three lowest, it lists the stop reasons the log records, such as yarn breaks, needle changes or a program change, and says where no reason was written. It builds the table as a spreadsheet.

    3. Result

      The meeting starts from one table of machines, output and recorded stop reasons, each figure traceable to its row.

Rejected rolls and idle machines: the published figures

The figures come from published studies and trade reports on knitting, each cited where it appears. They are industry figures, not Mavick results. In a pilot you take your own baseline on the KPIs below and measure against it.

  • Fewer rolls lost to faults found too late

    About 18%

    Knit fabric with major defects, which the study counts as not acceptable to customers: 4,149 m of 22,910 m inspected in one week on one Bangladesh knitting floor, before a 2022 Six Sigma project.

    How the agent helps

    The agent does not inspect fabric. When you ask, it counts the rolls with one outcome you name, such as rejected, for a defect type you name, and on one machine when your inspection table records the machine, from the uploaded rows. Ask again for the downgraded rolls. It builds the defect table as a spreadsheet and traces each roll to its machine and shift from the roll cards.

    Track: First-pass roll acceptance %, by machine

    Source: Textile & Leather Review, Implementation of the Six Sigma Methodology for Reducing Fabric Defects on the Knitting Production Floor: A Sustainable Approach for Knitting Industry (opens in a new tab), 2022.

  • When the knitting rate is below cost, every lost hour counts

    BDT 20–24/kg

    Cost of producing average single-jersey knitting in Bangladesh, while the rate paid for knitting the same fabric was BDT 12–15/kg, in a 2017 trade-press estimate by Textile Today Research.

    How the agent helps

    The agent reads no machine counters. From the shift sheet you upload, it totals the kg for a machine and shift you name from the rows and sets it against the target kg the sheet records. From the stop log, it lists the recorded reasons machine by machine, and counts the stops logged for one cause, such as yarn breaks, on one machine.

    Track: Machine efficiency % per circular machine, by shift

    Source: Textile Today, ‘Knitting’ growing fast with great prospect still without business viability (opens in a new tab), 2017.

  • Setting faults traced to the program and the machine

    30%

    Share of fabric defects traced to machine setting on one Bangladesh knitting production floor, in a 2022 Six Sigma study. Yarn quality accounted for 51%.

    How the agent helps

    The agent reads the record, not the machine. Ask about a machine, and it sets the last setting record beside the approved program and lists every value that differs or was never recorded. When your uploaded inspection table records the cause and the machine, it counts the rolls rejected for setting faults from the rows, one count for each machine you name.

    Track: Rolls rejected for setting faults, by machine and program

    Source: Textile & Leather Review, Implementation of the Six Sigma Methodology for Reducing Fabric Defects on the Knitting Production Floor: A Sustainable Approach for Knitting Industry (opens in a new tab), 2022.

What to baseline before a pilot

Measure each one before the pilot starts, then again at its end.

Knitting: the KPIs to baseline before a pilot, and why each matters
KPI to baselineWhy it matters
GSM within ±5% of targetThe playbook’s target is 100% of rolls. A share that slips points to yarn count or cam-setting drift, and GSM out of spec on a confirmed order means freeze, segregate and retest.
First-pass roll acceptance %The playbook targets 92% or more. Below 85%, it reads the cause as needle, oil or fly contamination.
Machine efficiency % on circular machinesThe playbook sets 80% or more and tracks it per machine. A floor average can hide the few machines losing the most hours.
Needle-line defect rate: needles failed per 1,000 hoursThe playbook sets a baseline for each machine and reads a spike against it as mechanical wear. The hours come from your shift sheet, when it records running time.
Width within ±1.5% on every rollThe playbook sets width within ±1.5% on every roll, checked beside GSM. Compare it with the width the buyer’s requirement states.

What the agent works from

Records it reads

  • Knitting program
  • GSM card
  • Roll card
  • Inspection report
  • Needle-change log
  • Shift production sheet
  • Stop log
  • Yarn issue sheet

In your ERP

Reads Odoo production orders and stock transfers, such as yarn issues, each with the time it was fetched. Drafts an order’s planned start or assignee, or a transfer note; an admin approves or declines each, and it runs only if approved. Yarn on hand, GSM and roll records come from an upload or a Google Sheet. The demo shows which yarn issues your Odoo holds.

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 the GSM sheet, roll log or needle-change log you keep in Google Sheets when you ask. Prepares a WhatsApp Business or Slack message to a shift supervisor, a row for your lot-hold sheet, or an email to the QA head through Gmail or Outlook. Each one waits for an admin to approve or decline it, and runs only if approved.

Tools named today: Gmail, Outlook, Google Sheets, Slack and WhatsApp Business.

Never made up

  • Machine settings: gauge, feeders, stitch length, tension
  • GSM readings, fabric width or shrinkage values
  • Production quantities or machine efficiency
  • Yarn lot numbers or blend composition

If it is not in your connected data, the agent says so.

Belongs to another department

What the Knitting playbook covers

  • SOP: every roll inspected, GSM at three points

    Inspect every roll for needle lines, drop stitches, holes, dirt, fly and oil. Check GSM at the head, middle and tail of each roll, and write the roll number and GSM range on the roll card. A roll-to-roll GSM drift above 5% means freeze and find the root cause.

  • Decision rule: GSM drifts on one machine in a shift

    Check three causes in order: the yarn supply (a cone change or a count change), the fabric tension (the take-up roller drifting), then the cam setting (an operator change). The playbook says most drift is yarn-related, so confirm the yarn before anyone touches the cams.

  • KPI: the ±5% GSM band, and when ±3% is possible

    The playbook’s standard is GSM within ±5% of target on every roll. Before you agree to a buyer’s ±3%, it asks for three things: cam-setting discipline, with operators certified on the gauge and count; GSM checks every 30 minutes, not once a shift; and yarn count CV% under 2%.

  • Heuristic: the needle that fails quietly

    A needle that snaps is easy to find. The one that hooks and only sometimes drops a stitch ruins a roll without anyone noticing. And if two machines on the same program show the same defect, the playbook says the cause is the program, not the machines.

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.

  • Approved knitting program

    What is the approved knitting program for this structure and GSM, and what settings are actually running on the machine?

  • GSM drift across the rolls

    Show the GSM readings across the rolls in this run and tell me whether the drift has gone outside tolerance.

  • Trace a needle-line defect

    Which machine produced the rolls with this needle-line defect, and when in the run did it start? Show the inspection records.

  • Machine-wise production this shift

    Show machine-wise production and efficiency for the last shift, and rank the machines by output against target.

  • Did a yarn lot change coincide

    Fabric appearance changed partway through this run. Did a yarn lot change coincide with it? Show the timings from the records.

  • Greige lot inspection summary

    Summarise the inspection results for this greige lot: defect types, quantities affected, and the overall grade.

  • Oil stains this week

    How much fabric was rejected or downgraded for oil stains this week, and which machines does it trace back to?

  • Machines idle and why

    Which knitting machines are idle right now, and what do the logs record as the reason for each?

  • Width and shrinkage on a quality

    What width and shrinkage were recorded for this quality, and how do they compare with the buyer’s requirement?

  • Yarn consumption against plan

    Show yarn consumption for this run against the planned consumption, and explain any variance from the records.

Knitting: questions before a demo

Our roll cards are hand-written. Can the agent read a photo of them?

No. A photo or a scanned card has no text for the agent to read. Once the roll records are typed into a Google Sheet, or an Excel or CSV file you upload, it answers from them, and from an uploaded Excel or CSV file it works out counts and GSM figures from the rows. Keep that table on the first sheet of the workbook. Knitting programs and SOPs that already exist as typed Word, Excel or PDF files (not scans) can be uploaded as they are.

Can it read stops, RPM or output straight from our circular machines?

No. Mavick does not connect to knitting machines, their counters or a machine-monitoring system. If that system exports stops and output to a file you upload, or you keep them in a Google Sheet, the agent reads them there, and from an uploaded export it totals a machine’s output or counts its stops from the rows. Knitting reads only production orders and stock transfers from Odoo, so stop logs never come from the ERP. Where a machine’s stops are missing from those records, the agent says so.

Can the agent change a knitting program or a stitch length?

No. It cannot send anything to a machine, and it does not write knitting programs into your ERP. It finds the approved program in your files and sets the setting record beside it. It does not work out a stitch length for a new GSM; that stays with your knitting master. The ERP changes that fit knitting are narrow: in Odoo, a production order’s planned start or assignee, or a note on a yarn transfer. Each waits for an admin to approve or decline it, and runs only if approved.

Which of our knitting records can it read from the ERP?

Two kinds, both from Odoo: production orders (order, product, quantity, status) and stock transfers, such as yarn issued to the floor (partner, status, scheduled date, source document). Odoo gives it no yarn on hand, and no other ERP gives Knitting any records. GSM cards, roll cards, stop logs and yarn stock come from a Google Sheet or an uploaded Excel or CSV file. For yarn consumption against plan, it sets the run’s yarn issue rows beside the planned quantity on the program sheet. From uploaded sheets with a run column, it totals the kilos issued to that run and the planned kilos, and gives the difference, citing both.

Will the knitting agent see dyeing’s complaint files or spinning’s yarn test reports?

No. Each department’s data stays inside it, and the knitting agent reads only what is connected or uploaded to Knitting. It cannot fetch from the dyeing or spinning departments. Asked about yarn count and quality parameters, it says Spinning owns them; asked about dyeing recipes or shade approval, it says Dyeing & Processing does. If a dyeing complaint or a yarn test report matters to knitting, upload a copy to the knitting department, or share it across the factory, and the agent reads it.

Can it tell us which yarn lot caused the barré?

No. It does not decide a cause. When you ask, it sets the yarn issue sheet beside the roll cards and inspection reports, so each roll’s yarn lot, machine, knitting time and recorded defect sit side by side, and it says where a roll’s lot was never written down. Barré can show only after dyeing, so upload the dye house’s complaint to Knitting and it reads that too. Which lot caused it is the knitting manager’s call; the playbook’s pre-run checklist asks for cones from a single lot so the question comes up less often.