The High-Stakes Bet on Automated Due Diligence

The Automation of Trust

The standard operating procedure for private equity has long involved hiring a small army of MBAs to grill a target company’s customers. These consultants charge six-figure sums to identify the risks hidden beneath a balance sheet. DiligenceSquared is betting that this entire layer of human capital is an unnecessary expense. By deploying AI-driven voice agents to conduct these interviews, the startup claims it can slash the cost of market research for mid-market acquisitions.

This shift represents a fundamental change in how capital is deployed. The official narrative suggests that AI can extract the same quality of insight as a human expert at a fraction of the cost. However, the move from human-led inquiry to automated polling ignores the nuance of high-stakes negotiation. A voice agent might follow a script, but it lacks the intuition to follow a thread when a customer’s tone shifts or when they omit a crucial detail about a product’s declining quality.

The Data Gap in the Script

DiligenceSquared positions its tech as a democratizing force for smaller firms that previously couldn't afford top-tier due diligence. They argue that volume is the new gold standard. If an AI can call 200 customers while a human team only manages 20, the resulting data set is theoretically more representative. But quantity rarely compensates for the loss of critical thinking. In the world of mergers and acquisitions, one disgruntled whistleblower is often more valuable than a hundred satisfied surveys.

The startup uses AI voice agents to conduct interviews with customers of the companies the PE firms are considering buying, aiming to make the research affordable for the mid-market.

The problem with making research affordable is that it often makes it commodity. When every firm uses the same automated tools to vet deals, the competitive advantage disappears. Furthermore, there is the issue of the 'uncanny valley.' Customers who realize they are talking to a bot may provide perfunctory, surface-level answers rather than the candid critiques necessary to stop a bad deal from closing.

Private equity firms are effectively outsourcing their skepticism to a black box. While the cost savings are immediate and undeniable, the long-term risk resides in what the AI fails to ask. If a voice agent cannot detect sarcasm, hesitation, or the subtle avoidance of a topic, the resulting report will be polished but potentially hollow. The cost of a failed $50 million acquisition far outweighs the savings found by cutting out a consulting firm.

Efficiency versus Efficacy

The tech stack behind these agents relies on large language models that are trained to be agreeable and helpful. In a due diligence context, you want the opposite: a persistent, investigative mindset. DiligenceSquared must prove that its agents can push back against optimistic projections and dig into the friction points of a business model. Without that friction, the 'research' is little more than an automated feedback loop.

We are seeing a trend where the infrastructure of finance is being rebuilt around speed. But speed is often the enemy of sound judgment. If these AI agents are simply checking boxes to satisfy a checklist, they are not performing due diligence; they are performing due diligence theater. Founders and investors should look closely at how these agents handle non-linear responses and whether they can actually identify a failing product before the check is signed.

The ultimate test for this model will be the first major write-down of an acquisition that was vetted exclusively by AI. Until then, the industry is trading depth for scale. The success of this venture depends entirely on whether an algorithm can develop the one trait that makes a great consultant: the ability to know when someone is lying to them.