Quick Answer: Propel One delivers the only truly agentic AI architecture in PLM because it's built on Salesforce's enterprise cloud platform with native access to reasoning engine, unified data, and embedded security, capabilities competitors cannot replicate through bolt-on AI features. While legacy vendors retrofit chatbots onto decades-old on-premise systems, Propel One's cloud-native foundation enables autonomous agents that orchestrate across PLM, QMS, and PIM with enterprise-grade governance, giving manufacturers faster development cycles and real-time impact analysis impossible on competing platforms.
The Architectural Divide: Platform AI vs. Bolt-On AI
By 2030, enterprises will increase AI investment by 64%, growing to $170 billion over the next five years.
But the PLM industry faces a fundamental split. There's a massive difference between having AI features and being built for AI.
Older PLM vendors are adding AI the only way their architecture allows: as an external layer on top of systems designed 20+ years ago. They're essentially building chatbots that query only their own product databases and maybe summarize help articles.
That’s where Propel One has a radical advantage: inheriting enterprise-grade AI infrastructure from the platform layer.
Propel CTO Kishore Subramanian explains that Agentforce is a full-featured AI platform... This includes a reasoning engine, integrated data lake, security context preservation, no-code tools such as Agent Builder and Prompt Builder, and pre-built orchestration capabilities.
The competitive implications are stark. With Propel One being built on top of Agentforce, users benefit from Salesforce developer priorities being laser-focused on an agent-centric experience.
Propel CEO Ross Meyercord and CTO Kishore Subramanian speaking with Industry Analyst Michael Finocchiaro for the "AI Across the Product Lifecycle" podcast.
What True Agentic Architecture Requires
The Five Non-Negotiable Components
Research from MIT Sloan reveals that even simple enterprise GenAI use cases require "20–30 elements" including LLMs, data platforms, security gateways, and orchestration. For true agentic AI—systems that autonomously plan, decide, and act—the requirements are even more stringent:
1. Planning and Reasoning Engine
Agents must decompose complex requests ("assess the cost impact of this change") into executable steps. This requires natural language processing combined with business logic understanding—not just keyword matching or document retrieval.
Propel One: Inherits Salesforce Agentforce reasoning engine natively
Competitors: Must build custom agentic layers on legacy databases designed for transactional queries, not semantic reasoning
2. Platform Unifying All Data Types
Agents need access to structured data (BOMs, change orders, approvals) and unstructured data (IP documents, CAD files, specifications, quality reports) in a single queryable layer. IBM's research on hybrid cloud and AI emphasizes this is "critical to adopting generative AI".
Propel One: Salesforce Data360 provides a native data lake that ingests from any source while preserving security contexts, meaning your data doesn’t leave the platform. Propel One is built right in.
Competitors: Data remains siloed across PLM, external documents, and integrated systems with no unified AI-queryable layer
3. Security Context Preservation
When AI agents access data, they must respect application-level permissions. A mechanical engineer shouldn't be able to ask an agent for financial data they don't have access to—even if that data exists in connected systems.
Propel One: Security models flow from application through data platform to AI layer automatically
Competitors: Must manually replicate permission structures into AI models or accept security gaps
4. API-Enabled Business Logic
Reading data is table stakes. Agents must take action: update records, trigger workflows, route approvals, generate reports. This requires comprehensive, well-documented APIs that agents can call programmatically.
Propel One: Every PLM function exposed as both UI and API; agents call the same services humans use
Competitors: Limited or poorly documented APIs; on-premise architectures weren't designed for programmatic access
5. 100% Scalable
AI workloads are unpredictable. When 50 users simultaneously ask agents to analyze supply chain impacts, the system needs elastic compute capacity that scales instantly then releases resources.
Propel One: Native cloud architecture with dynamic resource allocation
Competitors: On-premise systems require capacity planning and often fail under AI workload spikes
Why Legacy Architectures Cannot Compete
The On-Premise Anchor
Propel CEO Ross Meyercord, who implemented Windchill and other legacy PLM systems at Accenture, is blunt: "Companies still running 20+ year old legacy PLM or ERP systems are approaching a critical decision point."
The architectural constraints are insurmountable:
On-Premise PLM Systems:
- Store data in proprietary schemas optimized for 1990s transactional workflows
- Cannot integrate modern vector databases for semantic search
- Lack unified data lakes—documents live in file systems, metadata in SQL tables
- Require VPN access and cannot securely expose APIs to cloud AI services
- Force customers to self-host and manage AI infrastructure
According to Grand View Research, SaaS PLM now accounts for 74.1% of market revenue. PLM vendors like PTC and Siemens are caught in a trap: their installed base runs on-premise, their revenue depends on maintenance contracts for those installations, yet AI requires cloud infrastructure. Their "hybrid" strategies amount to bolting cloud AI onto on-premise data, which introduces latency, security vulnerabilities, and integration complexity.
The Single-Tenant Trap
Some vendors offer cloud deployments but in single-tenant architectures, essentially hosting separate instances for each customer. This model fails for AI because:
- No economy of scale: Each customer's AI infrastructure must be provisioned and managed separately
- Slower innovation: Updates require coordinating across hundreds of individual instances
- Limited data network effects: Agents cannot leverage anonymized cross-customer insights (e.g., supplier performance patterns)
- Higher costs: Customers pay for dedicated compute even when idle
Propel One's true multi-tenant SaaS architecture means every customer immediately benefits from platform improvements. When Salesforce enhances the reasoning engine or adds new AI models—delivering innovations as often as every two weeks, along with continuous enhancements to the AI—every Propel One customer inherits those capabilities automatically.
The Propel One Architectural Advantage: Real Examples
Expediting Change Orders
Propel One's expediting agent demonstrates capabilities impossible on legacy architectures:
What it does:
- Understands where each sits in workflow and what's blocking progress
- Proactively notifies stakeholders of required actions
- Takes autonomous action: updating fields, routing approvals, generating summaries
Impact Assessments
The agent automatically analyzes:
- Cost deltas between legacy and replacement parts
- Inventory implications for on-hand and in-transit materials
- Customer order impacts and revenue at risk
- Regulatory compliance requirements
- Supply chain alternatives and lead times
As Ross Meyercord describes Propel One's Impact Assessment agent: "The more data you feed it, the more robust and valuable its output becomes. There’s no single right answer, but this is where a voracious agent thrives: the more you give it, the more insight it delivers."
This reflects Tech-Clarity's research on digital thread orchestration, which emphasizes that PLM must orchestrate across heterogeneous systems to create "a rich AI/ML foundation".
Why competitors cannot replicate this:
Legacy vendors typically integrate PLM with ERP through batch file transfers or point-to-point middleware. Their "AI" might access PLM data, but cannot reach across to ERP, CRM, and supply chain systems in real-time with preserved security. The result: analysts still manually gather data from multiple systems, defeating the purpose of AI automation.
Propel One Agentic Architecture
As Kishore explains: "The agent actions are like building blocks... when the reasoning engine breaks out a query into step one, two, and three, [and] two and three are Propel specific, those skills are available, you can plug right in."
This modular approach is validated by research from Cambridge University on LLMs in engineering design, which maps systematic LLM task categories across engineering workflows, and arXiv's survey of LLMs for manufacturing, documenting multi-agent frameworks for design, quality, and supply chain.
Why this matters competitively:
Competitors building AI from scratch must develop these domain-specific capabilities themselves—a multi-year engineering effort. Propel One leverages both Salesforce's generic platform skills (data retrieval, workflow automation, natural language understanding) and adds PLM-specific skills on top. This dual-layer approach is unmatched in the industry.
Governance at Scale: The Enterprise Requirement
McKinsey's 2025 State of AI found that while 62% of organizations experiment with AI agents, high performers differentiate by redesigning workflows and mitigating risks around privacy, explainability, and regulatory compliance.
Propel One embeds governance mechanisms competitors bolt on as afterthoughts:
Knowledge Grounding
Agents are explicitly constrained to approved data sources. As Kishore notes: "You can ground [an agent] in certain knowledge... and as part of the instructions say just grounded here, do not go beyond this."
Mandatory Citations
Every AI-generated insight includes source references, enabling human verification and aligning with NIST AI Risk Management Framework trustworthiness principles.
Audit Trails
Complete logging of agent actions—what data was accessed, what decisions were made, what actions were taken—critical for FDA, ISO, and other regulatory compliance.
Continuous Coaching
Ross Meyercord shared an example where an agent pulled superseded documents: "We adjusted the prompt to only look at current documents... that took minutes to do." This rapid iteration is only possible when the AI layer integrates tightly with the application layer.
Gartner's AI TRiSM framework predicts enterprises applying these controls will improve decision accuracy by eliminating up to 80% of faulty information—a competitive advantage legacy architectures cannot match without years of governance infrastructure development.
The Proof: Customer Outcomes Competitors Cannot Match
Savant Systems integrated an acquired company's entire product catalog and made it orderable online on Day One of the acquisition close. Ross Meyercord: "Before Propel, they had no shot of doing that within months. But now, because all of that digital information is connected around product, they can take in that product information and push that out to the website in real time."
This demonstrates the compound advantage of Propel One's architecture:
- Connected data across PLM, e-commerce, and order management
- APIs that enable automated data synchronization
- AI agents that validate and enrich product data during migration
- Cloud scalability that handles data ingestion without performance degradation
Beta program engagement exceeded all expectations. Kishore Subramanian: "We would expect three or four people to show up to give us feedback. And there have been meetings where we've had more than 30 people show up to say, 'Hey, I have a use case... How can you solve it?'"
This enthusiasm reflects a fundamental truth: customers can immediately envision what's possible with true agentic AI once they experience it—because the architecture enables rather than constrains innovation.
The Strategic Verdict
BCG research on AI value shows leaders achieve 30% faster product development and up to 2% COGS reduction by concentrating AI in core functions like R&D and operations. But these outcomes require architectural foundations competitors lack:
As Ross Meyercord concludes: "Agentic AI and SaaS is a partnership imperative. Agents provide the intelligence layer; SaaS provides the operational foundation. Neither works effectively without the other in an enterprise environment."
For manufacturers evaluating PLM AI capabilities, the architectural differences aren't subtle; they're existential.
Propel One didn't bolt AI onto an old system. It was architected from day one to unleash what AI makes possible: autonomous agents that think, orchestrate, and act across the entire product lifecycle.
.png)
.png)
