CADDi: drawing retrieval can accelerate an RFQ without approving the quote

External-source research synthesis, not an OfferSignal client result. Based on the retained October 2026 research and source-specific dates.

A custom part rarely has one clean product record. Drawings, computer-aided design (CAD) files, bills of materials, quotes, supplier history, quality evidence and production notes may sit across enterprise systems, shared drives and email.

From procurement platform to manufacturing data layer

Founded in Tokyo in 2017 by Yushiro Kato and Aki Kobashi, CADDi began with custom-parts procurement and built internal tools for quoting, supplier selection, inspection and production. Its August 2026 relaunch positions Explorer and Agent as the current discovery and agent layer, while CADDi Quote handles requests for quotation. Drawer is legacy for existing customers, not offered to new customers; migration to Explorer is case by case. This is a documented strategic pivot, not evidence of a failed product or customer shutdown. 1

CADDi describes a Data Layer that ingests product-lifecycle management, enterprise resource planning, customer relationship management and internal-file data. It analyzes 2D/3D geometry, specifications, PDFs, presentations, bills of materials, images and audio. A Semantic Layer links parts, processes, defects, prices, suppliers and related records. Explorer searches and compares drawings, 3D CAD, quotes and quality records. Agent is described as able to analyze information, make decisions and take action within that context; the retained evidence does not establish autonomous purchasing, binding quote approval or machine control. 2

CADDi Quote covers supplier selection, RFQ distribution, response collection, comparison and award. It recommends suppliers using purchase history, similar drawings and quality or delivery performance, normalizes price, lead time and terms, and records communications and award rationale. This can prepare a sourcing decision, but does not replace engineering release, capacity confirmation or authority to issue a binding offer. 3

What named manufacturers reported

Unytite, a US manufacturer of bolts, nuts and structural fasteners, deployed Drawer across R&D, Quality, Sales and Production. CADDi reports nearly 60,000 drawings and production documents centralized, search in less than one minute, more than 90% lower engineering and production search time, and more than 200 hours saved. These are provider-reported results with no denominator, baseline period, control group or independent audit; they are directional customer claims, not general ROI. 7

Neaton Auto Products Manufacturing, a Tier-1 automotive supplier, loaded more than 140,000 documents from PLM, ERP and scanned handwritten records. CADDi reports active accounts rising from 30 to 85, described as a 183% increase; estimated drawing-search reduction of 40–70%; and QA search time falling from 3:17 to 1:30, approximately 50% faster. The estimates and timing are provider-originated, with no disclosed comparison sample. Neaton’s exploration of CADDi Quote was planned or ongoing, not a reported production result. 8

Amerequip, a custom equipment and hydraulics manufacturer serving OEMs including John Deere, Caterpillar and Oshkosh, used the system to search part numbers, drawing text and keywords, see assemblies using a component and detect duplicate parts. CADDi says this replaced ERP, cross-reference and drawing-server hopping; Design World corroborates the implementation. No quantified saving or before-and-after denominator is disclosed. 9 10

An illustrative governed RFQ pilot

This is an illustrative application, not a reported CADDi deployment. A manufacturer could select a versioned sample of drawings, quotes and quality records, then test retrieval against authoritative revision, unit, material grade, tolerance, process, tooling, inspection evidence and supplier capacity. Record search time, wrong-result rate, stale-revision retrieval and duplicate-part detection.

A historical quote is evidence, not a current price; a similar drawing is not proof of capacity. Keep role-based permissions, multi-factor authentication, single sign-on and audit logs around confidential drawings, and test malicious or stale retrieved instructions. Engineering approves technical fit; sourcing approves supplier selection and terms; authorised commercial owners approve any binding offer or purchase commitment. Public pricing is undisclosed and demo-based, so budget for extraction, integration, semantic mapping, migration, training and exception handling. 2 3 11

CADDi is a credible candidate for drawing and RFQ retrieval in custom manufacturing, with named deployments and measurable but provider-reported search outcomes. It should be evaluated as governed decision support—not yet as an autonomous purchasing agent, binding configure-price-quote system or universal ERP replacement. CADDi reports a $114 million Series D at a $1.2 billion valuation and $234 million total funding; earlier Series B and C disclosures should not be added to that current official total. 4 5 6

Sources and next reading

The CADDi AI Data Platform and CADDi Quote are public product overviews; commercial access and pricing require CADDi engagement. The Amerequip case study and Unytite outcome release are public customer materials, not self-serve applications.

Explore OfferSignal · Focused Business & AI Workshop · All research

References