Make vs Buy: Navigating AI Tool Development in Procurement

If you are banking on your traditional corporate procurement skills to carry you through the next two decades, it might be time for a reality check. In Episode 49 of the Proc & Roll podcast, hosts Natasha Gurevich and Zachary Bachir sat down with Jason Busch, the founder of Spend Matters and a highly provocative thinker in the procurement space.
Jason did not hold back, delivering a candid and urgent warning about how artificial intelligence is rewriting the rules of the corporate world. From the shrinking lifespan of legacy software to the end of traditional knowledge work, here are the core takeaways you need to navigate this massive shift.

1. The Three Tiers of AI: Beyond the Copilot

Many organizations are just starting to experiment with AI, but it is critical to understand the different levels of technology entering the workforce. Jason breaks this down into three distinct tiers:

• Copilots: These tools are not autonomous, but they solve human-based problems and have the potential to make humans three to five times more productive.

• Agents: The agentic layer involves a series of actions being taken on your behalf, acting as an extension of existing capabilities.

• AI Employees: This is the most disruptive tier. AI employees operate with a high degree of autonomy, can leverage existing technology, and even code on the fly. Crucially, Jason points out that AI employees actually have the potential to follow corporate policy and hard-coded constraints much better than humans do.

2. Legacy SaaS is Becoming a "Rolex"

The days of massive, multi-year software implementations are numbered. Jason argues that legacy SaaS has a shrinking "half-life" and is essentially becoming the corporate equivalent of a Rolex.

His advice for handling your current tech stack is bold:

• Treat every current SaaS investment as a sunk cost required simply to maintain the status quo.

• Simultaneously, spend equal time and energy actively trying to destroy that very same status quo.

• Start experimenting internally. Today, you can build 80% of a classic spend analysis solution using large language models for a fraction of the traditional cost.

3. The End of Traditional Knowledge Work

Perhaps the most jarring moment of the episode came when Jason discussed career advice for the next generation. He openly states that he would tell his kids to run away from procurement, supply chain, HR, and most of finance. Why? Because traditional, highly-credentialed knowledge work is exactly the type of labor most at risk of being replaced by AI. Instead of corporate roles, Jason suggests that true job security over the next 10 to 20 years will lie in manual trades and working with your hands. He highlights the massive data center build-out currently underway, noting that the country is short hundreds of thousands of electricians—a highly lucrative path in the AI era.

4. Build Your Digital Twin Before It’s Done to You

If you are waiting for a clear directive from the C-suite on how to implement AI, you are already falling behind. Jason warns that if procurement leaders do not aggressively adopt this technology, it will simply be forced upon them. Major consulting firms like McKinsey and Accenture are actively building agent ecosystems and selling them directly to the executives above you.
To future-proof your career, Jason recommends building a "digital twin" of your organization. Set up an AI-driven structure that works alongside your existing physical team to begin taking over tasks.
Start small: identify the tasks you hate doing—like delivering bad news to suppliers or syncing calendars—and find a way to automate them yourself.

The Last Mile of Humanity

While AI is poised to take over execution and data processing, Jason notes that the "last mile of humanity inside companies is critical thinking". As we enter this global AI race, cultivating the ability to argue multiple sides of an issue and exercise deep reasoning will be what sets human leaders apart.

Transcript: Proc-N-Roll | Make vs Buy: Navigating AI Tool Development in Procurement

Natasha: Welcome to Proc and Roll. Today we have an exceptional guest, Jason Busch, founder of Spend Matters. Jason, you recently posted some very provocative articles, including one comparing legacy procurement SaaS to a Rolex, and another saying you would tell your kids to run away from procurement. Can you share how your framework for thinking about this was shaped?

Jason: I started in consulting in the nineties and became interested in dynamic price mechanisms and game theory. I joined FreeMarkets because I wanted to build transparency and break down the massive information asymmetry in buyer-supplier relationships through reverse auctions. Today, I divide my focus into three main areas: proprietary data businesses that fuel AI, the AI-based layer itself, and the legacy SaaS world, which I believe has a rapidly shrinking "half-life".

Zachary: Looking at that AI layer, how do you see the difference between standard agents and the "AI employees" you are building? Are they going to run negotiations entirely on their own?

Jason: I think of it as a matrix of autonomy versus creativity and breadth. Co-pilots are not fully autonomous; they just make a human three to five times more productive. Agents take orchestrated actions on your behalf. But AI employees operate with incredible autonomy, speed, and depth. They can leverage existing technology and code on the fly. Interestingly, AI employees have the potential to follow hard-coded corporate policy and constraints much better than human workers, who rarely follow every single rule.

Natasha: In conversations with enterprises, when I ask who owns the AI strategy, they don't know if it's the CPO, the CFO, or the CIO. What guidance would you give to professionals to start embracing this technology?

Jason: If you're not doing it, it will be done to you. Consulting firms like McKinsey and Accenture are already building agent ecosystems, and they are selling them directly to the people above the procurement leaders. If you want to future-proof your career, try to set up a digital twin of your organization. Build a parallel structure that can work alongside your physical team and begin taking over tasks.

Zachary: The technology is moving incredibly fast. How should organizations navigate the "make versus buy" dilemma right now? Do you wait for a massive new platform, or start building it yourself?

Jason: I would literally treat every legacy SaaS investment right now as a sunk cost just to keep the status quo going, but I would spend equal time trying to destroy that status quo. I encourage piloting and failing fast. Have everyone on your team build an app to automate the tasks they hate doing, like telling suppliers bad news. You can build 80% of a classic spend analysis solution internally using large language models, and the cost to serve is a fraction of legacy tools.

Natasha: To make this completely practical for someone listening: where do people start today?

Jason: Find out what you have available internally, like Microsoft Copilot. I use everything: Claude for coding, OpenAI, and Grok, which is an insanely powerful research tool. Identify the monthly reports or communications that take up your time and automate them. More importantly, find the proprietary data you need to fuel that AI, whether it's commodity prices or even agricultural weather reports, because generic models need that data to be accurate.

Zachary: I agree with you that procurement is going to become a computer, essentially managed as a suite of AI applications. But what does this mean for the future of knowledge work overall?

Jason: The notion of procurement as knowledge work is completely changing. I wouldn't encourage my kids to go into traditional knowledge work like supply chain, HR, or most of finance because that is the work most at risk. For the next 10 to 20 years, until robotics and "world models"—which are AI models trained on real-world video data—take over physical tasks, working with your hands is incredibly valuable. For example, the massive data center build-out has created a shortage of hundreds of thousands of electricians, and they are making well into the six figures right now.

Natasha: If we are shifting away from traditional knowledge work, do schools need to focus more on things like critical thinking, writing, and logical reasoning?

Jason: Yes, but the classroom is changing. With AI tutors like Alpha Academy, a system can find a student's weakest link—like a math concept they missed in eighth grade—and drill them until they get it right. This allows students to learn the basics in greatly compressed time. Going forward, the last mile of humanity inside companies is critical thinking—specifically, the ability to argue multiple sides of the same argument.

Natasha: Finally, will this AI race impact how the world is run? Are we going to see democracy win, or will disparity increase?

Jason: We are in an existential race between historically Western values in AI and values based on central planning from countries like China. I think there will be a period of real disruption to the economy, and the pain will be unequally distributed among countries. However, long-term, I am incredibly optimistic because this technology changes everything and will increase the quality of life.

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