Jetpack Labs × UP&R

AI Quoting System — Process Architecture

End-to-end workflow from RFQ email intake through customer-facing quote delivery, replacing the manual Excel-based process.

Current Process
AI-Powered Process
Side-by-Side
Phase: Manual Quoting · Avg 15 hours per quote · 100% human effort
1

RFQ Email Received

Customer emails RFQ with attached print (PDF), material spec, and volume requirements to quoting@universalpolymer.com or directly to a sales rep.
Example: Bill Widmaier (All-State) → "Please quote P/N 92187688, EAU 2,000 pcs, PPAP Level III"
Manual Intake
2

Create Greenbook Entry

Estimator manually creates a row in the Greenbook Excel workbook: quote number, customer name, date received, part number, N/R flag, process type, salesman assignment.
2026-0142 | ALL-STATE INDUSTRIES | 2025-02-04 | 92187688 BRAKE CLUTCH STOP | N | R | JOE | BILL
Manual Entry
3

Create File Folder

VBA macro in Greenbook generates a folder on the file server. Estimator saves print, RFQ email, and cost sheet into the folder.
\\UPR-vm-fs01\Groups\Quoting\SALES\QUOTES\QUOTES 2026\ALL-STATE INDUSTRIES\2026-0142 92187688 BRAKE CLUTCH STOP
Semi-Automated
4

Read Engineering Print

Estimator reads the 2D PDF drawing to extract: cross-section geometry, dimensions (OD, ID, length), material callout, tolerances. Calculates cross-section area by hand or from experience.
Print says: OD 24.0mm, ID 19.1mm, Length 140mm, Material: 70±5 Neoprene → Area = π/4 × (0.945² - 0.752²) = ~0.258 in²
Expert Knowledge~30 min
5

Select Process Parameters

Based on 40 years of experience, estimator determines: compound number, extrusion rate (ft/min), number of strands, scrap %, MOQ, standard pack size, secondary operations needed, tooling requirements, packaging selection.
Compound: 0024005 | Rate: 25 ft/min | Strands: 1 | Scrap: 10% | MOQ: 2,100 | Std Pack: 350
Expert Knowledge~45 min
6

Enter Data into Cost Sheet

Estimator types all extracted values into the Excel cost sheet template (206 rows). Formulas auto-calculate material cost, labor, burden, scrap, packaging, and generate the margin tier table (15% through 50%).
Total Cost: $0.662/pc → @ 35% margin = $1.019 → @ 40% margin = $1.103 → Selected: 40%
Formulas RunManual Entry
7

Select Margin & Generate Quote

Estimator or sales manager selects margin percentage. Quote Form tab pulls data from the calculation sheet. Estimator prints/saves as PDF using Form 099 template.
Quote 2025-0142: 2,100 pcs @ $1.103 Ea | Tooling: $625 | Delivery: 6 wks ARO
Business Decision
8

Send Quote to Customer

Estimator emails the quote PDF to the customer. Updates Greenbook with date quoted and margin percentage. Saves all documents to file server folder.
Manual Send
Manual / Semi-Automated Step
Expert Knowledge Required
Phase: AI-Powered Quoting · Target <1 hour per quote · Human reviews, AI does the work
1

RFQ Email Auto-Intake

System monitors the quoting inbox. AI extracts: customer name, contact, part number, volume requirements, material spec, delivery needs, special requirements (PPAP, compliance). Matches customer against 951-record database.
Detected: Customer="All-State Industries" (96% match) | Contact="Bill Widmaier" | Part="92187688" | EAU=2,000 | PPAP Level III | REACH + RoHS
AI ExtractionAuto-Match
2

Auto-Create Greenbook Entry + File Structure

System assigns next quote number, creates Greenbook row with all metadata, generates the file server folder following UP&R naming conventions, and files all documents automatically.
Created: 2026-0142 | ALL-STATE INDUSTRIES | N | R | → \\UPR-vm-fs01\...\2026-0142 92187688 BRAKE CLUTCH STOP
Fully Automated
3

AI Print Analysis

Vision AI reads the engineering print to extract: cross-section geometry (OD, ID, wall thickness), part length, material callout, tolerance notes. For simple geometries (tubes, solid rounds, D-profiles), computes cross-section area. For complex profiles, flags for human review or uses SolidWorks model if available.
Extracted: OD=24.0mm (±1.0) | ID=19.1mm (±1.0) | Length=140mm (+4.0/-0.0) | Material="70±5 Neoprene" | Area=0.258 in² (confidence: 94%)
AI VisionConfidence Score
4

AI Parameter Selection

System matches material spec to compound database (520 compounds). Uses historical quote data to predict extrusion rate, strand count, scrap %, and secondary ops. Shows the 3 most similar historical quotes as reference.
Compound: 0024005 (match: "70±5 Neoprene" → 0024005, 98% confidence)
Rate: 25 ft/min (based on 47 similar profiles, σ=3.2)
Scrap: 10% (formula default, no override history for new parts)
Similar quotes: 2024-0891 (tube, 22mm OD), 2023-0445 (tube, 25mm OD), 2025-0023 (sleeve, Neoprene)
Historical ML520 Compounds4,000+ Quotes
5

Auto-Calculate Price

The exact same 206-row formula chain runs automatically. Material cost, scrap, setup, labor, burden, secondary ops, packaging — all computed with full audit trail. Margin tier table (15%–50%) generated. System suggests a margin based on customer history and market segment.
Material: $0.198 | Setup: $0.042 | Labor: $0.089 | Secondary: $0.00 | Packaging: $0.011 | Total Cost: $0.340
Suggested margin: 40% (All-State avg: 38%, distributor segment avg: 42%) → Price: $1.103/ea
Exact Formula MatchMargin Suggestion
6

Human Review & Approve

Estimator or sales manager reviews the AI-generated quote in a dashboard. Every AI decision is shown with confidence scores and reasoning. Side-by-side with similar historical quotes. One-click overrides for any parameter. Approve, edit, or send back for recalculation.
Review dashboard shows: 14 auto-filled fields, 12 high-confidence (>90%), 2 medium-confidence (cross-section: 94%, rate: 87%). Estimator adjusts rate from 25→28 ft/min based on tooling knowledge. Price recalculates in real-time.
Human DecisionAI Assists
7

Auto-Generate & Send Quote

System generates: (1) branded Form 099 PDF quote, (2) backup cost sheet with full formula audit trail, (3) updates Greenbook with date quoted and margin. Optionally auto-sends to customer via email template, or queues for manual send.
Generated: Quote_2026-0142.pdf | CS_2026-0142_92187688_brake_clutch_stop_ws.xlsx | Greenbook row updated | Email draft ready
Auto-Generated
8

Learning Feedback Loop

Every human edit during review feeds back to improve future predictions. Win/loss outcomes (tracked in CRM) refine margin optimization. Over time, confidence scores increase and override frequency decreases.
Estimator changed rate 25→28: system learns "Neoprene tubes with OD 20-30mm → rate bias +12%"
Quote won at 40% margin: reinforces distributor pricing model
Continuous LearningFeedback Loop
Fully Automated
AI-Assisted
Human Decision Point
Side-by-Side Comparison · Current vs AI-powered process
Step Current Process AI Process Impact
RFQ Intake Read email, identify customer, extract requirements manually Auto-extract from email + attachments, match to customer DB ~15 min → ~0 min
Greenbook Entry Manually type row into Excel workbook Auto-created with all fields populated ~5 min → ~0 min
File Organization VBA creates folder; manual file saving Folder created + all docs filed automatically ~5 min → ~0 min
Print Analysis Expert reads drawing, calculates geometry by hand AI vision extracts dims + computes area with confidence score ~30 min → ~30 sec
Parameter Selection 40 years of experience: rate, scrap, compound, tooling Historical ML suggests all params with similar-quote references ~45 min → ~15 sec
Cost Calculation Enter values into Excel; formulas run automatically Identical formulas run automatically with AI-selected inputs ~10 min → ~0 min
Margin Selection Human judgment based on customer, market, competition AI suggests margin; human confirms or overrides Human decision preserved
Quote Generation Fill in Form 099 template, save as PDF, email manually Auto-generate PDF + cost sheet; one-click send ~15 min → ~0 min
TOTAL TIME ~15 hours <15 minutes
(including human review)
60× faster
Requotes Find old cost sheet, update material price, recalculate Fully automatic — zero human input needed 100% automated
Knowledge Retention In Bill's head. Leaves when he retires June 30. Encoded in AI models + audit trail. Permanent. Risk eliminated
Consistency Varies by estimator (Bill vs Mark vs Z) Same logic every time. Overrides tracked and learned from. Standardized