AI-Native vs. AI-Washed: Don't Get Fooled by Fake Tech

The procurement technology space is currently flooded with promises of "AI-native" solutions, a trend that dominated the conversations and vendor booths at this year's DPW conference in Amsterdam. Beneath the marketing spin, many organizations are struggling to determine what is genuinely revolutionary and what is simply legacy software with an AI chatbot bolted on. Conrad and Zach cut through the hype to explore what the future of procurement technology actually looks like.  

AI-Native vs. AI-Washed

It is easy for software providers to claim their platforms are powered by AI, but true AI-native technology requires a fundamentally different architecture. Zach notes that many legacy technology platforms are simply bolting AI onto their existing structures. A genuinely AI-native system is built around a probabilistic inner core or model from day one, rather than trying to adapt older, rigid architectures to accommodate new tools. For procurement leaders evaluating new tech, it is critical to look past the banners and interrogate how the underlying architecture actually manages data and makes decisions.  

The Dangers of "Vibe Coding" in the Enterprise

The barrier to building software has never been lower. Conrad shared how he successfully "vibe coded" a fully functional global surf forecasting application in just a few hours using Claude and less than $100 in tokens. While building point solutions for internal or personal use is a great way to learn, doing this for enterprise applications carries massive risks.  

Conrad highlights a cautionary tale of a company that decided to vibe code a supplier onboarding application. Despite warnings, they pushed the AI-generated app into production, inadvertently opening a backdoor that led to a deep corporate hack. Without a robust operating model, proper IT governance, and an understanding of security and SOX compliance, citizen developers risk exposing sensitive company data.  

The End of Dashboards

The future user experience for procurement is natural language chat. Instead of forcing users to navigate complex, clunky interfaces and rigid ERP workflows like SAP, users will simply talk to their systems. An AI agent will reach down into the bedrock of the ERP, navigate the transaction codes, and proactively ask the user the right questions to populate a purchase requisition.  

In this future, traditional static dashboards become obsolete. Instead of clicking through pre-configured charts, procurement professionals will simply ask their AI agent to instantly generate an on-demand dashboard of outstanding POs.  

Job Replacement and the New AI Workforce

While AI will not replace every job, it will fundamentally shift how procurement professionals spend their time. Zach predicts that category managers, who currently spend heavy portions of their day on operational execution, will soon spend 90% of their time building relationships with suppliers and stakeholders, and only 10% reviewing the AI's execution.  

However, AI will absorb some roles entirely; Zach foresees a future where top-performing sales representatives could essentially be cloned into AI agents that read sentiment and handle responses. Consequently, a new class of jobs will emerge. Organizations will need professionals whose sole focus is to monitor, manage, update, and orchestrate these AI agents—essentially acting as managers for a digital workforce.  

The best way to prepare for this shift is to get hands-on with the technology. Just as messaging apps like WhatsApp seamlessly integrated into daily life, AI agents will soon become a natural, indispensable part of the enterprise workflow.  

Transcript: Proc-N-Roll | AI-Native vs. AI-Washed: Don't Get Fooled by Fake Tech

Conrad: Hello everybody, and welcome back to Proc and Roll, your guide to practical procurement. Today I'm coming to you from Amsterdam just coming out of DPW. Natasha, our good friend, is on a sailboat in the Mediterranean.  

Zach: I heard this year was one of their biggest years. Everyone has gone to DPW.  

Conrad: What was the main thing everyone's talking about at the conference? Like every conference this year, everyone's talking about AI. Every booth you went to had some banner claiming to be a native AI procurement platform. Are we in a place where this automation is really part of the tech stack people can buy today, or is it just marketing spin?  

Zach: You can buy some of it, but it really depends on your definition of what AI native means. For me, it's real autonomous procurement. On a technology front, it is a completely different architecture. Legacy platforms are bolting on AI a certain way, but if you're building a system around a model today, you're starting with a probabilistic inner core.  

Conrad: I've seen some companies go all the way back to the bedrock of their platform to keep up, but that puts a fair amount of risk into purchasing decisions right now. Many of these technologies are going to go away completely or have a complete transformation.  

Zach: It is very risky. Businesses need to catch up to how the technology works to ask good questions. An agent is like an employee you delegate tasks to, so you need to understand how it makes decisions and how it's governed. Companies doing use case identification exercises are missing the point. You need an internal operating system for your policies and what you are happy delegating to an AI before configuring the application layer on top.  

Conrad: I want to push back a little bit. I used Claude to build a tool that pulls global surf data and predicts waves. In less than 45 minutes, we had an application running, and I spent maybe four hours and under $100 in tokens. It's a point solution, and it works perfectly.  

Zach: That is awesome. Go experiment and vibe code, especially if it's an internal tool without external access. The risk is only high if there is some kind of access, and you only need one person from IT to sense check it.  

Conrad: Right. I was talking to a partner who mentioned a company that decided to vibe code a supplier onboarding application. He warned them not to, but they proceeded anyway. Sure enough, it opened up a hack door into their company, and a much deeper hack came into the business. Point solutions are great to learn, but you have to be careful when building for the enterprise.  

Zach: Exactly. The line for me becomes decision making. The minute you want an agent to start making decisions about which supplier it should award a contract to without human review, you need robust governance, compliance, and deterministic checks so they can't do anything stupid.  

Conrad: I keep presenting this concept of the consolidation of the procurement tech stack. A CPO recently asked an orchestration company to show him natural language procurement, and when they just showed last year's architecture, he basically threw them out of the room. The user experience for procurement of the future is natural language chat.  

Zach: I use Whisper on my computer, and just talking is 80% of my interaction with AI today.  

Conrad: If I can just talk to something, it can go all the way down to the bedrock of the ERP and say, "Hey SAP, open up this transaction and start a purchase requisition." The agent already knows how to navigate SAP's complicated codes and fields. There's going to be a massive compression inside this technology stack.  

Zach: I completely agree. You're going to have just systems of record and then agents that read and write into them. The buttons, bells, whistles, and dashboards are all on demand. An agent could just sit in a meeting, recommend starting a procurement process based on the transcription, and just present the PO to you for approval without you getting involved in the busy work.  

Conrad: I had breakfast with a former CIO, and he said this idea that "everybody is a builder" is a terrible idea in the enterprise. It creates freaking chaos. People are going to take sensitive financial information out of secure, SOX-tested systems and put it on their local machines. We both agreed that CIOs are going to get fired over this lack of governance.  

Zach: I agree with your CIO friend. Not everyone should build their own systems. You should have one joint operating model, and everyone builds on top of that while adhering to guardrails, security, and controls. You need rules about taking data out of systems or which domains agents are allowed to access.  

Conrad: I'm building automations internally to manage my operations, but it's not realistic for every single person in the enterprise to be an agent genius. I need my salespeople selling and my support reps answering calls. I don't see a future where every employee is building agents.  

Zach: The whole point of AI is that it frees you up. If a category manager spends 60% of their time on execution today, AI will handle that. In the future, they should spend 90% of their time building relationships and 10% reviewing what the AI has done.  

Conrad: You're not going to lose your job to AI, you'll lose it to someone who uses AI to do it better.  

Zach: Many jobs will be replaced by AI in 5 to 10 years, like some of the best sales reps being cloned into agents. However, a new class of jobs will be created around managing, maintaining, and updating these agentic platforms. Think of it like managing a team—you need someone looking after the agents, getting them onboarded, and making sure they're monitored.  

Conrad: If you're a leader in procurement, don't let this overwhelm you. By next Monday, you can have an agent up and running. Get into Claude, ask it how to do things, and just get started so you can intelligently form a vision for your tech stack before you buy into marketing hype.  

Zach: Don't be intimidated or nervous. Start at your own pace. Just like everyone uses WhatsApp today, everyone will be using AI tomorrow.  

Conrad: Thanks everybody for a great episode of Proc and Roll. Don't forget to like, comment, and share this episode with your friends out there.  

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