Building Digital Twins: How AI is Reshaping the Procurement Workforce

Welcome back to the Proc n Roll blog! In Episode 54, the "Three Amigos"—Conrad, Natasha, and Zach—reunite for a jam-packed session exploring the wild frontier of AI in the workplace, before diving deep into the architecture of procurement operating models.

The HR Debate: Are AI Tools "Employees" or "Agents"?

The episode kicks off with a fascinating debate currently happening across the tech and HR worlds. Natasha points out that many tech developers want to classify AI tools as "employees" to push the idea of hiring digital workers, while HR and talent acquisition leaders strongly insist on calling them "agents".  

Natasha firmly agrees with the HR perspective: you shouldn't equate agents to people. They are different constituencies that must be managed, structured, and rewarded differently.  

Meet "James": The AI Chief of Staff

Conrad introduces his own AI agent, "James," who acts as his Chief of Staff. Instead of just being a simple chatbot, James operates on a schedule and proactively manages workflows:  

• Meeting Notes & Action Items: Conrad uses a tool called Granola to record and transcribe meetings. James reviews these notes overnight, extracts the action items, and populates them into a Jira task board.  
• Team Accountability: James will actually open a Slack message with the relevant team members, paste the meeting notes, and ask for updates on their assigned tasks.  
• Proactive Scheduling: James reviews Conrad's calendar, drafts agendas for upcoming staff meetings, and even messages the team to update their travel plans.  

Conrad highlights how this feels like a full-circle moment for the corporate world. Decades ago, leaders had human travel agents and admins to manage their workflows. That eventually gave way to a frustrating era of "self-service" portals. Now, AI agents are bringing that white-glove, full-service support back to the individual employee.  

Security Guardrails: The "Mac Mini" Strategy

While this AI automation is incredible, it comes with risks. Conrad explains that giving an agent access to your web browser and Google account could theoretically give it access to your password manager or bank accounts. To mitigate this, many early adopters are isolating their agents on separate hardware, like a Mac Mini, to create physical boundary guardrails.  

However, Conrad predicts that within the next 12 to 18 months, major players like Google and Microsoft will release enterprise-safe environments that will unlock these agentic workflows safely for everyone.  

Architecting the Procurement Operating Model

In the second half of the episode, the team transitions from AI to the fundamental blueprint of a procurement organization: the Operating Model. Zach and Natasha break down how to structure and document the way your team delivers value.  

Choosing Your Archetype

Zach outlines several primary archetypes for organizing procurement work:  

• Centralized: All procurement routes through a single central function, often preferred in highly regulated industries like financial services.  
• Decentralized: Local decision-making is empowered to respond quickly to specific local demands and markets.  
• Center-Led: A hybrid approach where strategy and policy are set centrally, but execution is handled locally.  

Natasha provides a brilliant real-world comparison to illustrate this. A company like Salesforce, which operates primarily out of corporate offices without manufacturing or retail footprints, can easily use a Centralized model. However, a company like Nike, which has corporate offices, retail stores, manufacturing, and distribution centers globally, requires a Center-Led model. They need a global umbrella for policy, but local authority to manage things like regional payment term laws.  

The "Factory of Ideas" Framework

When building out the actual org chart, Natasha recommends a matrix approach. You place your spend categories (Direct, Indirect, Real Estate, etc.) as the vertical pillars, and then apply horizontal service layers across the entire enterprise:  

1. Category Management: The "factory of ideas." This requires deep subject matter expertise and human relationship building.  
2. Strategic Sourcing: The "factory of projects." This layer takes the ideas from category management and turns them into executed sourcing events.  
3. Contract Services: A dedicated legal procurement team that focuses solely on procurement contracts.  
4. Enterprise Programs: Organization-wide initiatives like Supplier Relationship Management (SRM), Centers of Excellence, and Sustainability.  

The Future: Building a "Digital Twin" of Your Org

The conversation circles back to AI with a mind-bending concept: creating a digital twin of your procurement organization.  

Instead of fighting for headcount, Natasha suggests that if you have 20 MRO buyers but actually need 30 to handle the volume, you can build 10 digital twins to augment your team. Zach takes this a step further, noting that because digital twins don't sleep, a digital category manager wouldn't just update a strategy document annually—it could continuously ingest data and update the strategy in real-time.

Transcript: Proc-N-Roll | Building Digital Twins: How AI is Reshaping the Procurement Workforce

Conrad: Welcome back to Proc n Roll, your guide to practical procurement! Today we have the "Three Amigos" back together—myself, Natasha, and Zach. We are diving into a jam-packed session exploring the wild frontier of AI in the workplace, specifically AI agents, before transitioning into the fundamental architecture of procurement operating models.   

Natasha: There is a fascinating debate happening right now. I recently had a conversation with Jason Busch where he insisted on naming his AI tools and calling them "employees." But later that same day, I watched interviews with HR and talent acquisition leaders who strongly insist on separating people and AI, calling the AI exclusively "agents." I completely agree with the HR perspective; you shouldn't equate agents to people. They are different constituencies and should be structured, managed, and rewarded entirely differently.   

Conrad: It’s incredible how far the technology has come. Remember those little Tamagotchi digital toys we had in school that would just beep to be fed? Now, I actually have an AI Chief of Staff named James. He has real personality—if I throw out some "aloha," he responds with surfer slang, and he can switch to Spanish instantly. But more importantly, he's incredibly functional and handles my accountability. I use a tool called Granola to record and take notes during my meetings. Overnight, James extracts all the action items from those notes, puts them into a Jira task board, and then proactively opens a Slack message with the relevant team members to follow up and demand updates. It feels like we are coming full circle. Decades ago, we had human travel agents and administrative assistants. Then we moved into a frustrating era of "self-service" portals. Now, AI is bringing that white-glove, full-service support back.   

Zach: That is a true agent because it runs on a schedule and plans its own work. I use an agent called Co-Work for my personal shopping and web browsing. However, you have to be very careful. To keep these tools secure and prevent them from accidentally wiring money from your bank account or messing with your passwords, you have to strictly specify what they can access. That's why many people initially isolated their agents on separate hardware, like a Mac Mini, to build a physical boundary.   

Conrad: Exactly. Though I predict that within 12 to 18 months, major players like Google and Microsoft will release enterprise-safe environments that unlock these workflows securely for everyone. But let's shift gears from technology to the blueprint of our business: the procurement operating model. How do we take the ingredients of people, process, and technology and document them into a working framework?   

Zach: You start by looking at operating model archetypes: centralized, decentralized, center-led, federated, or hybrid/outsourced. Different spend categories require different delivery models. For instance, buying global commodities is best centralized so you can control raw material purchasing and global shipping. Conversely, local indirect spend, like facilities management, needs to be decentralized so local teams can choose what fits their specific office needs. You document this structure through policies, org charts, job descriptions, and standard operating procedures.   

Natasha: To put that in perspective, look at the business itself. A company like Salesforce, which is primarily corporate offices, can easily run a centralized model. But a company like Nike, with corporate offices, retail stores, manufacturing, and global distribution centers, needs a center-led model. You have a global umbrella for policy, but local authority to comply with regional laws, like varying payment terms of 45 days in Europe versus other standards in Asia. When building the model, I approach it as a matrix. Spend categories act as vertical pillars. Then, you layer horizontal services that span the entire enterprise. The first layer is Category Management, which is the "factory of ideas" and requires deep human subject matter expertise. The next layer is Strategic Sourcing, the "factory of projects" that executes those ideas. Below that, you have Contract Services—a dedicated legal team for procurement—and Enterprise Programs like Supplier Relationship Management (SRM) or Sustainability.   

Conrad: How does AI impact these carefully structured operating models?   

Natasha: I am incredibly excited about the concept of digital twins. Our previous guest, Jason Busch, mentioned that building a digital twin of your organization is the ultimate starting point for breaking AI barriers. Practically, this means instead of fighting for more headcount, you use agents. If you have 20 human MRO buyers but need 30 to handle the volume without burning people out, you can simply build 10 digital twins to manage the extra load. It allows the operation to run smoothly and makes everyone's life more interesting.   

Zach: Exactly, and these digital twins don't need to work like humans. A human category manager might do a quarterly review and update their strategy document once a year. But a digital twin doesn't sleep. It can be switched on the whole time, continuously feeding on data, and updating the strategy in real-time. They can go a lot further than a human can.   

Conrad: This is all fascinating. As we wrap up, what is your top advice for a CPO who is starting to design or redesign their operating model today?   

Zach: You have to go back to the basics: what does procurement exist to do in your specific business? Do a diagnostic of your spend profile, ask your internal teams if they have the right tools, and evaluate your cost to serve. Identify where you are falling short, and tweak the model to maximize value and meet your customers where they are.   

Natasha: I agree, but my main advice is to learn the business first. You must understand the enterprise you are in completely before you build the operating model. That is the only way you can properly align procurement's priorities with the broader goals of the company.   

Conrad: Incredible insights as always! These aren't just documents; this is the true architecture and blueprint for implementing value and delivering services to the company. Thank you so much, Zach and Natasha. I learn something from you every time we talk. And to everyone listening, please don't forget to subscribe, like, comment, and share so we can continue bringing these practical insights to the world of procurement. See you next time!   

This transcript has been edited for clarity while maintaining all substantive content