AI in the Investment Office – Abby Barlow, Laura Hill, Brian Sugrue, Jenny Heller, John Lawrence, Matt Bank, Kristin Kallergis Rowland, Jon Webster (EP.515)

Episode Summary

Executive Summary: The episode compiles eight CIO perspectives on how AI is already reshaping investment offices: speeding meeting prep, document processing, reporting, and internal research, while still falling short on judgment and fully autonomous decision-making. Across firms, the common pattern is enterprise-grade tools, strong data organization, human verification, and a growing shift from external vendors to custom internal systems.

Main Topics: AI as an investment-office productivity engine (Priority: 5/5): CIOs use AI to summarize decks, prepare meetings, draft memos, process documents, and reduce administrative load across public and private markets. Human judgment remains the boundary (Priority: 5/5): Several speakers stress that AI should augment analysis, not replace core investment judgment, thesis formation, or accountability. Data organization and internal infrastructure (Priority: 5/5): The biggest enabler is not the model itself but clean, searchable, standardized data, knowledge layers, and internal workflows that let AI work effectively. Build vs. buy: internal tools are winning (Priority: 4/5): Many teams found vendor products underwhelming for nuanced workflows and increasingly prefer custom-built systems tailored to their data and process. AI as a thought partner and red team (Priority: 4/5): AI is used to challenge assumptions, compare historical manager behavior, generate counterarguments, and improve preparation before meetings or IC discussions. Trust, privacy, and verification discipline (Priority: 4/5): Enterprise security, sandboxing, no-training clauses, and human review are recurring requirements before teams allow AI into consequential workflows. From productivity to alpha generation (Priority: 5/5): The more ambitious firms are trying to move beyond efficiency gains toward better decisions, better risk sensing, and eventual alpha creation—though most say that is still emerging.

Key Arguments: AI is already saving meaningful time in meeting prep, document review, reporting, and financial analysis, but the gains are mostly incremental rather than transformative. The most effective use cases are structured, specific, and high-confidence tasks; AI struggles when the problem is broad, messy, or requires deep domain judgment. Human reviewers remain necessary because AI can produce plausible but wrong outputs, especially with numbers, complexity, and poorly constrained prompts. Internal knowledge systems and data lakes are prerequisites for serious AI use because unstructured, fragmented data limits model quality. Many organizations are moving from off-the-shelf AI products to internal development because third-party tools often fail to fit heterogeneous investment workflows. AI can improve decision quality by expanding the amount of information teams can process, but most CIOs still view it as support for judgment rather than a replacement for it. The next frontier is not just better documents or faster models, but better institutional memory, manager intelligence, risk sensing, and workflow continuity. The long-term test for AI is whether it makes the organization a better investor, not whether it simply makes work easier.

Data Points: Number of CIOs interviewed: 8 - Ted Sides asked eight chief investment officers how they are using AI today. Years Abby has been in role: 3 years - Abby Barlow said she started as CIO of the single-family office three years ago. Tracked changes in legal document: 185 - Abby used Claude to summarize a final-stage legal redline with 185 tracked changes. Time to build benchmarking app: about half a day - Abby said Claude helped turn an Excel-heavy benchmarking project into a usable app in half a day. Human review cadence: weekly - Laura Hill described weekly AI-generated document extraction emails for private markets statements. General GIST document workflow: capital calls and statements ingested - Advocate Health’s internal system extracts data from private markets statements and capital calls into a uniform template. Time saved on operations: roughly 30 hours - John Lawrence said Canoe saves about 30 hours for the operations team by processing incoming documents. AI podcast training input: about 2 hours - Rice used roughly two hours of prior podcast audio to generate an AI voice for a board presentation. Annual return impact target: 10 basis points - John Lawrence said a 10 bps increase in annual returns would be worth it. Internal code output: 30,000 lines per month - Matt Bank said GEM’s team generates about 30,000 lines of code per month, up 10x year over year. Team size on development: 3 developers - GEM shifted toward internally building tool sets with three full-time developers. Data history at JPMorgan: over 40 years - KK Rowland noted the complexity of ingesting more than four decades of data into the AI stack. Engineering support at JPMorgan: over 50 engineers - KK Rowland said she oversees a team with over 50 engineers supporting $250 billion of assets. Client asset base mentioned: $250 billion - KK Rowland described AI deployment across $250 billion of alternative assets under her watch. Private bank AUM referenced: $500 billion - KK Rowland said $250 billion of the private bank’s $500 billion AUM is in alternatives. CPP Investments AUM: $580 billion US - John Webster referenced CPP Investments’ scale as the largest Maple 8 pension fund. Beneficiaries served: 22 million Canadians - John Webster framed AI’s role as improving decisions on behalf of 22 million Canadians. Workflow acceleration target: 45 days to 4–5 days - John Webster said AI is helping move delivery from 45 days down to 4–5 days. Tool adoption time frame: 20 hours - John Webster said people become much more effective after roughly 20 hours working with AI. Alternative investment AUM under KK Rowland: $250 billion - KK Rowland described the scale of the private bank alternatives platform she oversees.

Pivotal Quotes: "We will use AI to augment human expertise, accelerate routine work, and improve decision quality while preserving human judgment, confidentiality, and accountability." — Laura Hill: Advocate Health’s one-sentence AI soundbite, capturing the episode’s core boundary between automation and judgment. "It’s not a better PowerPoint, it’s not a better model, it’s not a better memo. AI can do all of those things. But it’s speed to decision, differentiated relationships, differentiated viewpoint." — Laura Hill: Laura explains that AI’s value lies in freeing time for edge and judgment, not in polishing outputs. "The test for AI is: does it make us a better investor? Better underwriting, better risk-taking, better decisions on behalf of 22 million Canadians." — John Webster: CPP Investments’ framing of AI as an investment-quality question rather than a technology experiment.

Implications: AI in investment offices is moving from novelty to operating infrastructure. Winners will likely combine clean data, enterprise security, and disciplined human oversight to turn AI into a durable productivity and decision edge.