AI sits above your PLM. For Centric PLM and PTC FlexPLM, the supplied evidence supports an orchestration model that reads and writes through connectors and APIs while PLM stays the source of truth. It does not confirm a working connector for either product or explain its mechanics. You need proof of that connection before you buy.
Systems of record
Whitepaper 03: The Enterprise Trust Blueprint places the AI layer above PLM, ERP, PIM and DAM, with exports in.glb,.png and.pdf. Those systems keep their role as records of truth. The proposed model avoids replacement and addresses version drift, such as a spreadsheet measurement edit that never reaches the factory tech pack. Governed records also reduce copy-paste steps where unapproved changes can enter. The paper names no supported products.
PLM strengths
State of AI Fashion 2026 says Centric, PTC FlexPLM and Backbone lead on governance and audit trails, deep ERP costing links, multi-brand scale, supplier networks and change-management maturity. This is a qualitative judgment rather than a benchmark test. Keep those controls in your buying criteria: the report positions production AI as an addition to PLM, so draft speed alone cannot settle the choice.
Connection scope
Whitepaper 06: The Agentic AI Rollout Playbook maps system types rather than product connectors. Trend work needs e-commerce analytics, POS, returns reasons and your catalog; social signals, competitor assortment feeds and search trends can accelerate it. Design needs DAM, brand style guidance, print libraries and color standards, with prior-season performance and a material library as further inputs.
Merchandising needs planning, inventory, historical sales and pricing rules. Costing, allocation, replenishment and forecasting tools can extend that workflow. AI tech packs need PLM, a tech-pack repository, measurement and material libraries. A supplier portal, approvals workflow and QA defect logs can help carry work toward factory use. Use this map to check the scope you need before comparing vendors.
Factory handoff
The Machine-Readable Brand models meaning retained as 100 percent in the design system, 60 percent after PDF export, 35 percent in email and about 15 percent in a factory-chat screenshot. No sample or method is stated, so treat these figures as an illustration rather than measured loss. They frame a useful check: can your factory get the approved record without reconstructing it from copies?
Buying criteria
The F* Word generates tech-pack drafts in 8-10 minutes and moodboards. Tech-pack release needs technical-designer review. For your software choice, ask for proof of the PLM connection and approved handoff. These sources name no winner and leave Centric and FlexPLM connection details unverified.
Sources: The F* Word; Whitepaper 03: The Enterprise Trust Blueprint; State of AI Fashion 2026; Whitepaper 06: The Agentic AI Rollout Playbook; The Machine-Readable Brand.
Get The F* Word workflow insights in your inbox.