There is a dynamic playing out in LP-GP relationships right now that most fund managers have not yet processed fully.
Institutional investors are deploying their own AI tools to ingest and analyse what their managers send them - most of which still arrives as a PDF. The LP's AI system is not waiting for a human analyst to open the document and read it. It is processing it the moment it arrives, extracting data points, cross-referencing answers against previous submissions, flagging inconsistencies and generating a risk-scored summary for the investment committee.
The manager who spent three weeks carefully crafting a DDQ response - pulling answers from last year's document, updating the numbers manually, running it through compliance review - may be submitting a beautifully formatted PDF that an algorithm will assess in sixty seconds. And that algorithm will find the inconsistency on page 23 that the manager's own team missed.
What the data is showing
Private Equity International's LP Perspectives 2026 Survey found that 47% of LPs are closely monitoring GP AI adoption. While a third view AI positively, 46% hold mixed views owing to risk concerns. What this headline figure does not capture is the speed at which LPs are simultaneously deploying AI in their own investment and due diligence workflows.
LPs are explicitly asking about AI adoption during due diligence calls, incorporating technology questions into DDQs, and directing capital preferentially toward GPs demonstrating operational innovation. For GPs, this creates both risk and opportunity. Forward-thinking firms that embrace AI will differentiate themselves in fundraising and potentially command better economics, while firms clinging to legacy processes will face increasingly difficult LP conversations.
"LPs are not waiting for their managers to catch up. Institutional investors are deploying their own AI tools to ingest and analyse what their managers send them - most of which still arrives as a PDF."
- Quadsight, Why AI Adoption in Investor Relations Keeps Stalling, April 2026Very few alternative managers have an AI policy that would survive scrutiny in an operational due diligence review. The gap between what LPs can now do analytically with the documents they receive and the quality of documentation most GPs are still submitting is widening - and managers who do not close it will start experiencing the consequences in allocation decisions.
The three-part problem
The first problem is structural inconsistency. When a manager completes DDQs for multiple LPs simultaneously using shared drives and email threads, version control breaks down. Answer A in the pension fund DDQ differs subtly from answer A in the endowment DDQ because different team members completed different sections under deadline pressure. An LP using AI to cross-reference answers across multiple submissions from the same manager will surface these inconsistencies immediately. A human analyst reviewing one document in isolation would likely miss them.
The second problem is data structure. Delivering clean, structured, timely data is not just a service improvement for fund managers. It is a chance to shape the narrative, deepen the relationship, and demonstrate operational credibility at exactly the moment it matters. An AI system ingesting a badly formatted PDF with tables broken across page breaks, inconsistent date formats and manually typed numbers cannot extract clean data points. The LP's AI system will flag the document as low quality before a human has opened it.
The average PE fund responds to 150+ DDQs annually during fundraising, each averaging 250 questions across 21 categories. Response windows have compressed to approximately 7 days. Maintaining consistency across 150 responses to 250 questions per response - under a 7-day window - using a manual process based on a shared Word document is not operationally viable. Errors are not an outcome of carelessness. They are a structural certainty of the process.
The third problem is the audit trail. When an LP's compliance team asks to see the basis for a specific answer submitted eight months ago, the manager using email-based DDQ management cannot reconstruct who provided that answer, when it was approved and whether the same answer was given to other LPs. The manager using centralised, version-controlled DDQ infrastructure can produce that audit trail in minutes.
What the new standard looks like
The managers navigating this well have made a specific infrastructure change. They have moved their DDQ management off email and shared drives and onto a centralised platform where every answer is version-controlled, every change is timestamped, and every response pulls from a single authoritative source. When the compliance policy is updated, every pending and future DDQ response reflects that update automatically.
- Single source of truth - one authoritative answer to each question, updated centrally, pulled into every DDQ response automatically regardless of which team member is completing it
- Full audit trail - timestamped record of who provided what answer when, making regulatory review and LP follow-up questions answerable in minutes rather than days
- Structured data output - responses formatted and structured for clean AI extraction, ensuring the LP's system generates an accurate risk score rather than flagging poor data quality
- Cross-consistency checking - platform flags where answers in the current DDQ differ from answers given in previous submissions before the document leaves the building
- Parallel processing - multiple team members contribute simultaneously without version conflict, meeting the 7-day response window without cutting corners on accuracy
If you are reviewing a manager's DDQ using AI to assist with analysis and consistency checking, the quality of your output depends directly on the quality of the input. A DDQ submitted as a well-structured, consistently formatted document with clean data points will yield far better AI-assisted analysis than the same information buried in a manually completed PDF. The question to add to your next ODD interview is not just whether the manager uses AI - it is whether their DDQ infrastructure produces output that your AI tools can actually work with.
The allocator perspective
For ODD teams conducting manager reviews, the volume of documentation being processed is increasing faster than headcount. AI tools saving 30-40 hours per due diligence process are not a luxury for these teams - they are a structural requirement for maintaining analytical rigour across a growing manager universe without proportionally growing the team.
The practical implication is that ODD workflows need the same centralised infrastructure upgrade that the best managers are implementing on the submission side. A cloud-based ODD platform that centralises incoming DDQ responses, extracts and structures data points consistently, tracks manager documentation over time and flags changes between submissions is becoming table stakes for institutional allocators managing large manager universes.
AlternativeSoft's Due Diligence Exchange provides the centralised cloud infrastructure that both sides of this equation need. For managers, it is a single source of truth for every DDQ response - version-controlled, audit-trailed and structured for clean data extraction. For allocators, it provides a platform to manage the full ODD workflow with consistent data quality across every manager in the universe.
When the LP's AI starts reading before the IR team has finished writing, the infrastructure you are running matters as much as the answers you are giving.