Where a style goes wrong in the design room
Each problem starts in a record the design room keeps: the tech pack version, the marker report, the pattern approval register, the DST file. When those drift apart, the cost turns up later, in the sample room, at cutting or on the embroidery machine.
Bulk arrives narrower, and the costed marker no longer holds
The marker is laid for the order’s size and colour ratio at the planned cuttable width. Then bulk fabric arrives narrower, a one-way or check fabric needs another marker type, or the ratio is heavy on one size. The efficiency and consumption that were costed no longer hold, and nobody can say quickly which styles sit under the playbook’s 85% for bottoms or 80% for tops.
90%
Estimated share of all cutting-room loss that sat inside the marker itself, in a 2026 study of denim trousers by researchers at National Textile University, Faisalabad (1,728 observations). It names marker width among the risk factors.
Source: Springer Nature (Environment, Development and Sustainability), Determining the cutting wastage and risk factors involved in the manufacturing of denim trousers – a step towards sustainable apparel industry (opens in a new tab), 2026.
A style on its fourth pattern, and no one is sure which set is approved
Grade rules differ by buyer and by POM. A fit comment changes the base pattern, the graded set is re-issued, and the register still shows the old one as approved. A junior’s pattern goes out without the senior review the playbook insists on. Styles wait for approval while sample dates pass, and every extra round adds sampling cost the factory often carries.
41.2%
Share of suppliers industry-wide who reported being paid for sample development at all, in Better Buying’s 2025 supplier ratings of buyers in apparel, footwear and household textiles.
Source: Cascale (Better Buying), Better Buying Purchasing Practices Index Report (BBPPI) 2025 (opens in a new tab), 2025.
Tech pack v3 in the CAD room, v2 on the cutting table
The pattern master grades from one version, the sample room sews from the next, and the fit comment that moved a POM sits in an email nobody filed with the tech pack. An inch spec converted to centimetres by hand adds its own error; the playbook names conversion as the most common cause of a mis-cut. The mismatch shows at cutting, and the playbook escalates it to Merchandising and the plant head the same shift.
A logo that looks clean on screen and puckers on the fabric
Artwork is digitised into a DST file and the preview looks clean. On a light jersey, a fill denser than the playbook’s reference (about 1,500–2,000 stitches per square inch, satin rows 0.35–0.45 mm apart) puckers the fabric and breaks needles. If bulk starts before a stitched first piece is approved, or on a file edited after it, the embroidery section stitches the mistake into every panel.
Tech pack, pattern set, marker, DST: one answer for each
It works from files your design room keeps: tech packs, CAD marker reports, grade-rule tables, the DST register. Each answer names the file, version or rows it used, and any email, message or sheet row it drafts waits for an admin to approve or decline.
Tech pack versions on file, compared with their sources
Reads your documents
Upload each tech pack version and the buyer’s comment emails. Ask for a POM and its tolerance, and the agent quotes the latest version and names the file. Ask what changed since v2, and it lists the POM, tolerance, trim and label changes it finds, quoting each version; check the list against the files before you lock. With your metric spec sheet beside the buyer’s inch tech pack, it quotes the POM from both.
Ask“What do the tech packs on file give for front rise on style 5108, and did it change between v2 and v3?”
Marker efficiency by style and width, built into a weekly sheet
Creates files
Upload one marker log as Excel or CSV, a row per marker: style, cuttable width, ratio, efficiency, consumption. Ask for the average, lowest or highest efficiency for a style or a width, and the agent calculates it from the log’s rows and cites them. Add a column against the playbook’s target, such as “Below target: yes”, and it counts those too. It builds the weekly marker sheet. It does not lay or nest a marker.
Ask“Make this week’s marker sheet with each marker’s style, cuttable width and efficiency, and count the markers marked below target.”
The approved pattern set and grade rules, from your register
Answers with sources
Keep the pattern and marker approval register as a Google Sheet or an Excel file: style, set version, grade rule, drafted by, reviewed by, status and dates. The agent says which set and grading are approved for a style and who reviewed them, and lists the styles still waiting, oldest first. From an uploaded register it counts a style’s revisions and sets the count against the playbook’s limit of two.
Ask“Which pattern set and grade rules are approved for style 5108, and who reviewed them?”
The approved embroidery file, stitch count and placement
Answers with sources
From your DST register and the embroidery production worksheet exported as PDF, the agent gives the file name, version, stitch count, thread chart and placement on record for a design, with its approval date. Ask whether tomorrow’s worksheet matches the approval, and it compares the two records value by value. It cannot open a DST file, count stitches or check density.
Ask“Which DST file, stitch count and placement are approved for logo LG-208, and does tomorrow’s worksheet match?”
A tech-pack query drafted within the playbook’s 48 hours
Acts in your tools
Ask whether the brief answers a construction question. The agent quotes the clause with its file and version, or says it is not there. For a gap, it writes the query email to the buyer’s technical team, or to your merchandiser to send on, to go from Gmail or Outlook, and a new row for your query log in Google Sheets. Both wait in Mavick; an admin approves or declines each one, and each runs only if approved.
Ask“Does the brief give the stitches per inch for the inseam on style 5108? If not, draft the query to the buyer.”
A gap in the brief, narrower bulk fabric and the pre-cut check
A construction question the brief does not answer
Pattern master
The sample room asks whether the back yoke on a new shirt is single or double needle, a day after the tech pack arrived.
Asks
“Does the tech pack or any buyer email for style 7731 say whether the back yoke is single or double needle? If not, draft the query.”
The agent
Searches the tech pack versions and buyer comment emails uploaded to the department. It finds no answer, quotes the two clauses that come closest with their file and version, and says it is a gap. It drafts the query email to the buyer’s technical team with those clauses, and a row for the query log in Google Sheets with the date raised.
Approval
The email and the log row wait for the design head, as department admin, who approves or declines each one. They are sent and written only if approved, and the decision stays in the audit trail.
Result
If the head approves inside the playbook’s 48 hours, the email runs from the connected mailbox to the buyer with the clauses that were checked, and the log row records the date it was raised.
Bulk fabric arrives narrower than planned
CAD marker maker
Bulk denim for a trouser order arrives narrower than the cuttable width the marker was laid for.
Asks
“Bulk for style 5108 came in at 142 cm cuttable. List the markers on file for it with width, ratio, efficiency and consumption, and say if any was laid at 142 cm.”
The agent
Reads the marker log uploaded as one Excel file. It counts the markers for style 5108 from its rows, lists them with width, ratio, efficiency and consumption as recorded, citing each row, and counts those laid at 142 cm. If none was, it says so and quotes the efficiency recorded at the old width, so the new marker has a figure to beat.
Result
The marker maker knows whether a marker already exists for the new width or a new one must be laid in CAD, before cutting asks for it.
The pre-cut check before a style goes to cutting
Design and CAD head
A style is due in the cutting room on Monday, and the buyer’s last comment came in on Thursday.
Asks
“Run the pre-cut checklist for style 5108 and show the record behind each line.”
The agent
Reads the tech pack versions, the approval register, the marker report, the DST register and the first-piece record. It builds a document from the playbook’s five pre-cut checks: tech pack locked, pattern set graded and verified, marker laid and efficiency confirmed, DST file and version on record matching the approval (it compares what is written; it does not hash the file), first piece approved.
Result
A checklist the head can review before the style is cut, each line citing its record or saying “not on record”, with each revision of the document kept.
What a wrong spec and a lost centimetre of marker cost
These are published figures on late spec changes, cut-and-sew waste and embroidery defects, each cited at its source. None is a Mavick result. A pilot measures your own design room against the scorecard below, starting from today’s numbers.
More of the fabric you buy ends up in garments
18%
Share of material wasted in cut-and-sew (640 of about 3,500 tonnes) for one outdoor brand’s mostly polyester and nylon production in Vietnam, Fall and Spring 2022. A university master’s project; all cut-and-sew waste, not marker loss alone.
How the agent helps
The agent does not nest markers or choose a ratio. From the marker log you upload, one row per marker, it calculates the average efficiency for a style or a width from the rows, and counts the markers the log marks as below target. The CAD room then sees which styles to re-pattern and which ratios to raise with Merchandising, as the playbook advises.
Track: Marker efficiency % by style and cuttable width
Source: UC Santa Barbara, Bren School of Environmental Science & Management, Examining Cut-and-Sew Textile Waste within the Apparel Supply Chain (opens in a new tab), 2024.
Brief gaps raised early, and late changes listed by version
Over 30%
Share of general managers of 75 Cambodian garment factories in Better Work, surveyed 2015–2018, who called a buyer’s change to technical requirements after production began a problem. Cambodia only.
How the agent helps
Asked whether the brief answers a question, the agent quotes the clause or says it is not there, and drafts the query with the clauses it checked; it is sent only if an admin approves. Given a revised tech pack, it lists the changes it finds against the version on file. From an uploaded query log it counts a buyer’s open queries.
Track: Tech-pack queries sent back to the buyer within 48 hours %
Source: Better Work (ILO and IFC), Sourcing Practices in the Garment Industry: The Root Cause for Poor Working Conditions (opens in a new tab), 2022.
Embroidery first pieces approved, and bulk run to them
18.46%
Share of defects that were wrong thread colour in one run of 4,902 knitted garments in an industrial embroidery section, where 1–3% of pieces were defective. The authors link it to poor preparation or checks before the machine starts.
How the agent helps
The agent cannot open a DST file or check density. It answers from your DST register and embroidery worksheets: the file, version, stitch count, thread chart and placement on record for the approved first piece. Ask, and it says whether a value on a run’s worksheet differs from the approval or was never written down.
Track: DST first-piece approval rate
Source: MDPI (Textiles), Analysis of Quality and Production Efficiency in Industrial Embroidery (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 |
|---|---|
| Marker efficiency % by style and cuttable width | The playbook targets 85% or more for bottoms and 80% or more for tops. Below that it looks at three causes: an awkward size ratio, pattern pieces that do not nest, and marker-making skill. |
| Tech-pack queries sent back to the buyer within 48 hours % | The playbook targets 95% or better. Raise the query early: if the buyer then cannot answer within 48 hours, the playbook says the ship-date pressure is on them, not you. |
| DST first-piece approval rate | The playbook targets 90% or better, and says never to run bulk on a DST whose stitched sample has not been checked. |
| Pattern revisions per style | The playbook targets two or fewer. It reads more than that as a sign of an unclear brief, which points back to the tech-pack queries. |
| Days a style waits for pattern or marker approval | Not a playbook target, so baseline your own. A style that waits holds up the sample room and pushes back the cut date. |
What the agent works from
Records it reads
- Tech pack (every version)
- Buyer comment emails
- Grade-rule table for each pattern set
- Marker log or CAD marker export (one row per marker)
- Pattern and marker approval register
- Tech-pack query log
- Embroidery DST register
- Embroidery production worksheet
In your ERP
Reads only Odoo sales orders (order, customer, total, status, order date), Odoo invoices and SAP S/4HANA sales orders; it writes nothing back. None carries order lines, styles, sizes, patterns, markers, tech packs or DST files: those come from uploads and Google Sheets. The demo shows which of these sales orders and invoices your ERP connection 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
When you ask, it searches the design room’s Gmail and reads its Google Sheets registers; from Outlook it reads recent emails. It prepares the tech-pack query email, a query-log row, a status change in the pattern approval sheet, or a Slack or WhatsApp Business message to the embroidery supervisor. An admin approves or declines each; it runs only if approved.
Tools named today: Gmail, Outlook, Google Sheets, Slack and WhatsApp Business.
Never made up
- Measurements, tolerances or grading rules
- Marker efficiency percentages or fabric consumption figures
- Tech-pack versions or buyer approvals
- Embroidery stitch counts or file versions
If it is not in your connected data, the agent says so.
Belongs to another department
- Buyer approval correspondence and order termsMerchandising
- Cutting, and how markers are used on the cutting floorGarmenting & Stitching
- Fabric consumption costingCosting & Pricing
- Fabric development and trialsR&D & Product Development
What the Design & CAD playbook covers
SOP: reading the buyer’s tech pack
A buyer’s tech pack often runs to about 40 pages. Cross-check fabric construction, GSM, BOM, measurements, trims, label pack, AQL and packing. Send every missing or ambiguous field back to the buyer within 48 hours, and lock the tech-pack version once the buyer confirms.
Decision rule: a revised tech pack after the PO
When the buyer sends a new version after PO confirmation, apply it only to the production still to come. Pieces already cut or sewn to the earlier version stay as they are, and you tell the buyer in writing. The playbook warns against quiet re-work: it is expensive, and it tells the buyer that a confirmed tech pack is only a sketch, not a spec.
KPI: two pattern revisions per style, at most
The playbook’s target is two revisions or fewer for each style. More than that, it says, usually means the brief was unclear. Its rule for the pattern room has no exceptions: a pattern drafted by a junior is reviewed by a senior.
Heuristic: a stitched sample tells the truth
Converting artwork to a DST file loses fine detail, and the needle sets the limit. For a complex logo, show the buyer a digital simulation and a stitched sample before you lock the file, and never run bulk on an unverified DST. A DST whose file size or hash no longer matches the approved record has been edited, so it needs a new approval.
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.
What the current tech pack specifies
What does the current tech pack specify for this style — construction, measurements, trims and finishing? Cite the version.
What changed between versions
Compare the last two tech pack versions for this style and list exactly what changed.
Approved pattern and grading
Which pattern set and size grading are approved for this style, and what grading rules were applied?
Marker efficiency for a style
What marker efficiency was achieved for this style at this fabric width, and what does the marker file record?
Current embroidery file
Which embroidery file and placement are current for this design, and what is the stitch count?
Does the brief answer this question
Does the buyer’s brief answer this construction question, or is it a gap? Quote the relevant part of the brief.
Tech pack change summary
Prepare a change summary for this style from the tech pack versions on file, in date order.
Styles awaiting pattern approval
Which styles are awaiting pattern or marker approval, and how long has each been waiting?
Fabric consumption per garment
What is the fabric consumption per garment for this style from the marker on record, and how does it compare with the costing assumption?
Measurement points flagged by QC
Which measurement points on this style has QC flagged, and what does the tech pack specify for each?
Design & CAD: questions before a demo
Can the agent open our Gerber, Lectra or Optitex files?
No. It cannot read native pattern or marker files, or DXF exports of them, and it does not draft, grade or nest. Export what your CAD records as text instead: the marker report, the grade-rule table, the size spec. Upload them as PDF, Excel or CSV, up to 10 MB each, and the agent quotes width, ratio, efficiency and consumption as recorded. In an Excel file, the first sheet is the table it calculates from.
Can it check stitch density or digitise our artwork?
No. The agent does not digitise, simulate stitching or measure density, and it cannot open a DST file, which holds stitch data rather than text. It reads the embroidery production worksheet your software exports as PDF, and your DST register: stitch count, size, colour sequence and placement as written. It can quote the playbook’s density reference for a solid fill, but the stitched first piece stays the check, as the playbook says.
Will it tell us when the buyer sends a new tech pack version?
No. It does not watch your inbox or your folders, and it sends no alerts. When you upload the new version and ask, it compares it with the one before and lists the changes it finds, quoting both files. If the change came only in an email, upload that email or connect the design room’s Gmail, which the agent searches when you ask. From Outlook it reads only recent emails. A scanned tech pack has no text to read, so ask the buyer for the original PDF or the spec sheet in Excel.
Can it set QC’s measurement findings against the tech pack?
Only from a report the design department holds. If QC’s measurement report for the style is uploaded to the department as PDF or Excel, the agent sets each POM it lists beside the tech pack’s spec and tolerance, quoting both files. It does not see records kept inside QC’s own department, and it does not judge the garment; it quotes what each report records.
Can it compare the marker’s consumption with our costing?
Only with figures the design department holds. It reads consumption per garment from your marker reports as recorded. If the consumption sheet you issued to costing is uploaded to the design department, the agent sets the two side by side and cites both. It cannot open the sheets uploaded to Costing & Pricing. An organization admin can upload a copy to the whole factory, and the agent then reads that copy. Asked to cost the fabric, it says that department owns fabric consumption costing.
Our pattern approval register is on paper. Can the agent still help?
Only once it is typed in. The agent cannot read a paper register, and a photo or scan of one has no text to read. The ERP connection does not cover it either: it reads no pattern, marker or DST records. Move the register into a Google Sheet, which the agent reads whenever someone asks, or an Excel file you upload, from which it counts a style’s revisions. Start with the tech packs and marker reports you already have as files.
Bring Design & CAD. 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)