“Why the Dawn of Autonomous Procurement Is Still a Mirage”

1. Introduction

Jeff Morris Jr.’s article “When Software Buys Software“ paints a vivid picture: enterprise procurement is shifting to algorithmic, impersonal processes where “the best product wins, not the best schmooze.” He argues that the era of steak dinners and wine-country retreats is waning, replaced by AI agents autonomously evaluating and purchasing software.

It’s a compelling vision—but is it reflective of reality? In this response, I argue that this premise overstates the maturity, adoption, and practicality of “software agents buying software.” In truth:

• Human relationships, complex negotiations, trust, and regulatory nuances still drive enterprise software decisions.
  
• AI agents remain nascent, experimental, and far from replacing the human touch, especially in B2B contexts.
  
• Real-world trends suggest hybrid models—not wholesale automation—are the dominant trajectory.
  

Let’s unpack these points in detail.

2. AI Agents Are Real—but Still Experimental

There’s no question that AI agents are gaining attention. ARK Invest notes a spike in interest as firms like Microsoft, Salesforce, and UiPath rebrand around AI agents Ark Invest. Goldman Sachs estimates that AI agents could significantly expand the software market—possibly growing the customer-service software segment by 20–45% by 2030 Goldman Sachs. Gartner and IDC also forecast that by 2028, one-third of enterprise applications could include AI agents, with IDC imagining domain‑specific agent fleets replacing entire apps within a decade Insight Partners.

However, these remain futures—not today’s commonplace.

• A Wikipedia article on “Intelligent Agent” notes that “as of April 2025… few real‑world applications of AI agents” exist, and most companies are still in the experimentation phase Wikipedia.
  
• A Reuters analysis underscores that while AI agents are gaining traction, their business models remain hazy, and their implementations are complex and lacking proven returns Reuters.
  

Bottom line: AI agents show promise—but are not yet driving algorithmic procurement at scale.

3. The “Buy vs Build” Shift: Emerging—but Limited

Another trend often confused with algorithmic procurement is the rising use of AI in software development itself, reversing the traditional “buy vs build” calculus.

• Business Insider reports that AI-coding tools—like Bolt, Replit, and Cursor—empower non-engineers to build internal tools with natural‑language prompts. Some firms are replacing SaaS tools with these internally developed AI-powered apps Business Insider.
  

While this signifies innovation, it’s not the same as autonomous agents choosing commercial software. It’s organizations using AI to build, not buy. And even here, concerns persist—e.g., maintenance, complexity, and reliability Business Insider.

4. Relationships Still Matter – Human Trust Isn’t Dead

Corporate procurement remains a deeply human, relational, and often political process:

• Public filings or strategic deals—like Salesforce’s $8B acquisition of Informatica—demonstrate how critical negotiation, integration, and trust remain in enterprise software decisions The Wall Street Journal.
  
• ServiceNow’s acquisition of Moveworks (for $2.85B) signals that enterprise AI products are built via human-led M&A strategies, not algorithmic pick-and-buy agents Barron’s.
  

Morris’s assertion that “the best product wins, not the best schmooze” oversimplifies reality. Relationship-building, vendor reputation, SLAs, legal negotiations, service quality, and compliance—all human-centered factors—still dominate enterprise deals.

5. Limitations, Risks, and Governance of AI Agents

Even the most sophisticated AI agents face major hurdles:

• Reuters warns of unclear business models for AI agents and highlights complexity, risk, and uncertain returns Market Report Analytics
  
• Wikipedia details a raft of concerns: liability, security vulnerabilities, ethical issues, hallucinations, lack of explainability, regulatory ambiguity, job displacement, and more Wikipedia.
  
• A law‑based study on “Responsible AI Agents” recognizes these agents’ power but emphasizes that humans must remain responsible—regulation, oversight, and alignment are essential, not optional arXiv.
  

These concerns suggest that enterprises are far from handing procurement over to unchecked AI logic.

6. Hybrid Models Prevail—not Full Automation

Instead of a digital agent purchasing software independently, the current trajectory is toward hybrid systems:

• AI-assisted interfaces—like IBM’s watsonx Orchestrate—automate workflows and assist with tasks like onboarding or procurement, but are not autonomous buyers Standard Beagle.
  
• Coupa’s AI‑powered procurement tools now include an “autonomous AI agent network,” but it’s designed to recommend and support transactions—not to finalize them without human supervision Wikipedia.
  
• ServiceNow projects its AI agents (like Now Assist) will grow dramatically, but acknowledges that much of the revenue will still come from core subscriptions and acquisitions—not autonomous deals Investors+1.
  

AI is becoming a co-pilot in procurement processes—not a full replacement for humans.

7. Sales Still Drive Software—Despite AI Hype

The enterprise sales model remains resilient:

• The Wall Street Journal argues that despite AI hype, the dynasty of enterprise software remains strong: AI tools can codify some workflows, but firms like Workday and Salesforce are integrating—not erasing—this infrastructure; full AI-driven displacement remains impractical The Wall Street Journal.
  
• Private equity acquisitions of companies like Verint and Calabrio point to traditional financial/legal due diligence—but these deals are not made by AI agents buying software from each other Reuters.
  

8. Summary: Why the Premise Fails Today

9. Conclusion: Let’s Be Realistic—Not Pessimistic

Jeff Morris Jr.’s article offers a provocative future: lean, efficient, agent-driven software procurement. But present reality doesn’t align:

• AI agents are enhancing—but not replacing—the procurement process.
  
• Relationship-driven governance, corporate politics, compliance, and human judgment remain central.
  
• Risks—ethical, operational, legal, and financial—demand that human oversight remain integral.
  

Rather than proclaiming the obsolescence of enterprise sales, a more balanced view accepts AI as a force multiplier, not a sovereign agent. The future of procurement will likely be augmented, not fully automated.

References & Further Reading

• ARK Invest on AI agents and enterprise spending shifts newinternet.tech+7newinternet.tech+7newinternet.tech+7ARK Invest Europe+3Klover+3Market Report Analytics+3TechRadarWikipedia+4Ark Invest+4Reuters+4
  
• Gartner/IDC forecasts and IBM’s watsonx Orchestrate examples Standard Beagle
  
• Goldman Sachs’ projections for software market expansion via agents Goldman Sachs
  
• Wikipedia on intelligent agents and risks thereof Wikipedia
  
• Reuters: business model uncertainty around AI agents Reuters
  
• Responsible AI Agents (Desai & Riedl) study on governance and accountability arXiv
  
• Business Insider: AI coding tools shifting buy/build dynamics Business Insider
  
• WSJ: enterprise software resilient despite AI disruption The Wall Street Journal
  
• Salesforce and ServiceNow AI strategic moves and acquisitions Investors+2Barron’s+2
  
• Wikipedia/enterprise acquisitions (e.g., Synopsys–Ansys) for human-led M&A context Wikipedia
  

In short: The narrative of “software buying software” is an enticing vision—but still future-tense. Today’s procurement remains human-centric, complex, and relationship-driven. Until AI agents can reliably navigate legal, financial, ethical, and technical hurdles, humans will hold the steering wheel.