Polymr

The seam between engineering and the shop floor.

Discrete shops live at the seam between engineering and purchasing. BOMs change weekly, drawing packs arrive in three to five formats. The buyer-engineer-planner handoff is the death zone. Two of every ten revisions miss an affected open PO and surface 3 to 5 weeks later as scrap, a wrong-rev shipment, or a vendor invoice that no longer matches. Polymr closes that seam without rewriting PDM, ERP, or CAD.

Built for the discrete shape.

Polymr parses engineering drawings (PDF, STEP, IGES, DWG, SLDPRT) into routing-master proposals. Geometry, tolerances, callouts, and finishing notes resolve into a step sequence with work-centre suggestions and cycle estimates. The proposal lands on the process engineer queue with the prior-revision routing pinned side by side so the diff is visible at sign-off.

A BOM revision diff propagates into every open PO, every work order, and every quoted sales order touching the affected part. The two-of-ten miss rate collapses because nothing is buried in a separate engineering change notice queue. Make-vs-buy classification runs per line at quote time, against the live vendor catalog, and the engineer reads it before the quote sends.

A work order is a step sequence with a clock on every operation.

WO-1124 runs the brake-hub assembly through five work centres. Each step carries its status and owner. When step 030 runs one minute over standard on a 280-unit batch, the four-and-a-half-hour shift overrun surfaces against today, not against the W30 close.

WO-1124Brake hub assy PMR-4031 rev v6, qty 280
due 2026-06-09 14:00
  1. 010 . WC-204
    cut + chamfer billet
    done
    .
  2. 020 . WC-211
    CNC face + bore
    done
    .
  3. 030 . WC-211
    tap + cross-drill
    running
    .
  4. 040 . WC-308
    zinc plate + bake
    queued
    .
  5. 050 . WC-501
    assemble + final QC
    queued
Step 030 running over standard by 1 min on a 280-unit batch. Surfaces against today, not month-end.

WC-211 ran 118 percent on Wednesday. The plan locked Friday afternoon.

The work-centre heatmap is the planner's first surface every Friday at 15:00. Each cell shows percent loaded against available capacity for the upcoming week. Anything over 100 percent triggers a re-sequence proposal that the planner can approve in three clicks, or override with a documented exception.

Work centre capacity, W23 forecastpercent loaded vs available
MonTueWedThuFri
WC-204
78
84
102
96
71
WC-211
88
94
118
112
84
WC-308
62
71
81
79
58
WC-411
70
78
90
88
72
WC-501
54
60
66
64
51
WC-211 Wednesday at 118 percent. The 18-point overload is what triggered the W23 re-sequence on Friday afternoon.

Every PDM check-in walks the open POs and WOs.

The drawing-rev list is the engineering-to-purchasing seam rendered in one screen. SolidWorks PDM releases rev v6 of the brake hub at 16:18. Polymr computes the BOM diff, walks the open POs and WOs against v5, and lands three items on the buyer's queue at 16:19. Within minutes one is auto-approved (tolerance-only change), one goes to the buyer, one bounces back to engineering because the CAD attachment was missing.

Drawing revisions, last 24 hPDM change-doc trail
  • PMR-4031
    v5 to v6, geom change
    3 POs, 2 WOs
    awaiting buyer
  • PMR-4124
    v3 to v4, tolerance only
    1 PO
    auto-approved
  • PMR-3811
    v8 to v9, finish change
    4 POs, 1 WO
    rejected, CAD missing
Every check-in walks the open POs and WOs. The buyer sees three items, not three hundred emails.

Seven failure modes a BOM-driven discrete shop lives with.

Each is what a discrete operations director will name first when asked where the engineering-to-purchasing week disappears. PMR-* are internal placeholders; PO-##### and V-### are typical identifiers.

  • PDM-to-inbox lag
    Engineering posts rev v6 in PDM on Friday, the buyer team sees it the following Wednesday

    PMR-4031 brake-hub assembly steps from rev v5 to rev v6 at 16:18 Friday via SolidWorks PDM. The change notice goes to the engineering distribution list. Purchasing is on the customer-relations distribution list. The change is not surfaced against open PO-84179, PO-84212, or WO-1124 until the buyer manually sweeps the PDM exception report at 09:40 Wednesday. Five business days of releases ship against the old rev in between.

  • Drawing-pack disagreement
    A drawing pack carries M8 bolt qty 4 on the PDF and qty 6 in the supplier spec email

    The PMR-4031-A assembly drawing v6 calls out four M8x24 bolts in the title block. The supplier spec email forwarded from sales engineering says six. The buyer accepts the email number because it arrived last. The discrepancy is caught at receiving when V-244 ships six, the line consumes four, and two surplus bolts per assembly accumulate in WC-204 over the next 320 units.

  • Indefinite review status
    1,099 of 1,284 active BOMs carry at least one component stuck in "review" status indefinitely

    A component lands in "review" status the first time engineering substitutes it on a one-off job, the first time a supplier ships an out-of-spec replacement, or the first time finance flags a cost-basis question. None of those clearing paths have a workflow owner. The review queue grows by 8–14 entries a week. Active BOMs touch on average 1.3 review-status components.

  • Cost-basis chain breaks on rev
    A substitute-component chain breaks on revision and standard cost lies for the next 30 days

    When PMR-4406 v3 supersedes v2 mid-quarter, the v2 cost-basis chain (raw → forging → finishing → freight → landed) gets archived rather than carried forward. The v3 standard cost defaults to last-quoted instead of last-landed for the first 30 days post-revision, and downstream assembly margins on three product families read 3–5 points high on the operations dashboard.

  • Tolerance-standard mismatch
    One drawing pack uses ISO 2768, another uses ASME Y14.5, supplier RFQ scores compare apples to oranges

    PMR-4218 v4 was drawn against ISO 2768 medium for general tolerances. The customer-supplied mating spec for the next assembly was drawn against ASME Y14.5 with explicit datum references. Three suppliers quote the part. Two interpret per ISO, one interprets per ASME, and their unit prices span 31% because their inspection burden differs by an order of magnitude. The buyer scores them on price alone and picks the cheapest.

  • Alternate-part email never reaches the master
    A supplier-emailed alternate part offers a 9% saving and never enters the cost-basis chain

    V-244 emails the buyer suggesting STD-M824-B in place of STD-M824 at a 9% unit-cost discount with equivalent A325 spec. The buyer replies "thanks, will consider" and moves on. The alternate never gets added to the item master, never gets a preferred-vendor flag, never enters the next quarter's quote roll-up. The same email arrives again six months later, and the same exchange repeats.

  • Customer rev audit trail
    A customer asks which rev shipped against PO-84179 and the answer takes three days to assemble

    The customer-quality team at the receiving end of PO-84179 has flagged a fitment issue and asks which BOM rev was on the shipped units. The answer requires cross-referencing the shipping date, the WO that fed the shipment, the BOM rev active on the WO release date, and the actual lots consumed (which may have been substitutes). That cross-reference lives in three systems and one engineer's memory. The reply takes three days.

What this looked like at a sub-assembly contract manufacturer.

Anonymized engagement
Operations director, contract manufacturer of mechanical sub-assemblies
Situation
Drawings landed from three customer engineering teams across the week. Each revision triggered a full BOM re-type into the ERP, an RFQ packet build, and a sweep of open POs for downstream impact.
What was breaking
Of every ten revisions, two missed an affected open PO. Those misses surfaced 3–5 weeks later as the wrong rev shipped to a customer, a scrap event on the floor, or a vendor invoice that no longer matched the PO.
  • BOM extraction + RFQ
  • Engineering revisions
  • Quote-to-procure
Outcome · 6 weeks
100%
Revision-to-PO coverage
was 78%+22 pts
Illustrative, reflects this specific deployment. Outcomes vary by plant, stack, and scope.