A Guide for Procurement Leaders: Why Building Your Own Sourcing AI Isn't Worth the Wait

In my recent conversations with customers and prospects, a common theme is emerging. Several have recently told me that when transforming procurement with AI, the biggest risk is moving too slowly. In a recent webinar, for example, our customers Invesco and Bristol Myers Squibb both urged their peers in procurement to 'just get started'. 

So, what causes delays? One area where we see teams getting stuck is around the classic 'build vs. buy' question. To address this, we've launched a new white paper, "Build vs. Buy: The CPO's Guide to Evaluating Enterprise Sourcing Technology," which walks you through the decision-making process.

Drawing on real customer deployments, the paper expands on why general AI models aren't enough to handle complex sourcing and negotiation; the high failure rate and long timelines of building internal tools; and the hidden costs and risks of maintaining custom-built solutions.

Here's the spoiler: if you want the fastest, safest, and most cost-effective path, you need a platform built specifically for procurement, not a general-purpose AI.

The Limitations and Risks of LLMs

The main approach weighed up in the paper team is whether to use generic large language models (LLMs) to run certain procurement tasks or develop in-house systems. While these excel at providing conversational interface and generating documents, they can't run sourcing events, evaluate suppliers, or negotiate deals. They lack the specialized logic to understand requirements and manage competitive bids. On top of that, building your own tool typically takes about 17 months. And while you're waiting, competitors using specialized tech are pulling ahead.

Developing an internal solution also brings hidden costs and a high risk of failure. At least half of these projects never even reach production because of bad data, weak controls, or spiraling costs. Even when they do work, they require a permanent engineering team to maintain the tech. Every month spent building your own tool is a month of missed opportunities and unrealized savings.

The Fastest Path to Transformation

The white paper advises companies to focus on specialized platforms that deliver immediate, proven value. These use years of accumulated knowledge from thousands of real-world business events to create clear, structured processes for comparing suppliers. They can interpret needs, find the right suppliers, and handle the entire process from start to finish. They even manage simultaneous negotiations with every invited supplier, handling all the back-and-forth until they provide a final, decision-ready recommendation.

Purpose-built enterprise platforms deploy and deliver value within 30 days of the first sourcing event, requiring no prior data preparation. Customers produce 60 percent to 90 percent faster sourcing cycle times and 10 percent to 20 percent cost savings. 

To understand the complete total cost of ownership and evaluate the capabilities required to run successful events, download the white paper "Build vs. Buy: The CPO's Guide to Evaluating Enterprise Sourcing Technology" from the Globality website.