The introduction of AI in M&A due diligence is no academic hypothesis: in 2026 it is already operational practice in major international firms and Italian top-tier advisors, with documented efficiency metrics. The transformation operates on four concrete dimensions of the due diligence process: document analysis speed, depth of anomaly identification, process management cost, quality of final reporting for the decision-maker.
Document analysis speed is the most tangible delta. A typical mid-market VDR contains 800-2,500 documents between commercial contracts, financial statements, disputes, patents, certifications. Traditional due diligence with a 3-5 analyst team takes 4-6 weeks for first-pass review. With AI specialized in contract review and document intelligence — tools like Kira, Luminance, Harvey, integrated into ad hoc workflows — first-pass time drops to 7-12 days with more complete coverage: instead of sampling representative contracts, the system processes 100% of documents and autonomously identifies those requiring deep human review.
Depth of anomaly identification is the qualitative dimension that surprises most. AI identifies patterns that junior humans systematically miss: cross-references between separate commercial contracts that together create change-of-control issues, anomalies in contract renewal dates that concentrate risk in the first 12 months post-closing, exclusivity clauses incompatible between supplier and customer contracts, systematic omissions in regulatory certifications. These are the “surprises” that historically emerge only 3-6 months post-closing, when the contract is already signed and the buyer discovers undisclosed obligations.
Process management cost reduces in inverse proportion to dossier complexity: small standard deals see modest savings (10-20% on team cost), complex deals with thousands of documents see significant savings (35-50%). For the seller this translates into two benefits: more contained DD budget on buy-side if supporting competing offers, and reduced buyer sensitivity to their own DD cost (which becomes less of an argument for price discount).
Quality of final reporting is the least emphasized factor but perhaps the most impactful. AI allows automatically generating structured DD reports with direct citations to source documents, quantitative risk dashboards of identified risks, sensitivity scenarios on contractual terms identified as problematic. The senior advisor spends less time formatting and more time interpreting strategically — added value shifts toward judgment where humans are irreplaceable.
Two critical warnings, however. First: AI does not replace the senior — it empowers them. All decisions on materiality, strategic relevance, risk calibration remain and must remain human. Entrusting final decision to AI output is serious error and source of liability. Second: AI quality depends heavily on customization to sector and specific case. A generic system produces mediocre output; a system customized to transaction type, contract language, local regulatory framework produces superior results. In the Italian context this means adaptation to Civil Code peculiarities, sector regulations (industry, banks, energy, healthcare), and Italian contractual practice — area where advisors with Italian M&A experience have advantage over imported generic tools.
