Sep 14, 2026 · 1h 5m · capital-allocators
AI in the Investment Office – Abby Barlow, Laura Hill, Brian Sugrue, Jenny Heller, John Lawrence, Matt Bank, Kristin Kallergis Rowland, Jon Webster (EP.515)
Ted Seides · 11m spoken John Webster · 7m spoken Abby Barlow · 6m spoken KK Rowland · 6m spoken Matt Bank · 6m spoken Brian Chagru · 6m spoken Jenny Heller · 5m spoken John Lawrence · 4m spoken Laura Hill · 4m spoken Audio Track · 0s spoken
0:00 / 0:00
⌖ your search result is the highlighted band (1:02:32–1:03:01). Playback starts there
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Host Ted Seides interviews eight chief investment officers and institutional leaders to examine how artificial intelligence is transforming investment offices across organization sizes. The discussions highlight practical applications in operational automation, thesis red-teaming, and institutional memory codification, while defining the essential boundaries of human judgment in capital allocation.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Ted holds 19.5% of the talking time here. How this is scored →
Ted as informed peer 2.9 Guest teaching 5.8 Guest disagreement 1.3 Ted pushing back 1.0
speaking balance: gold is Ted, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 1:02:34 Rejecting technology as a lasting source of edge
Webster firmly rejects the common industry premise that AI technology itself confers alpha, stating that because frontier tools are universally accessible, lasting advantage resides solely in organizational EQ and trust.
Hardest push from Ted ▶ 11:30 Ted interrogating manager confidentiality in public LLMsTed directly challenges Barlow on the compliance hazard of feeding confidential GP fund documents and pitch decks into external AI models.
Biggest teaching moment ▶ 1:02:45 Applying Christensen's profit conservation theory to asset managementWebster educates Ted on Clayton Christensen's law of conservation of attractive profits, demonstrating that commoditizing analytical IQ inevitably shifts scarce value to relational EQ.
Ted holds their own ▶ 47:49 Ted framing the spectrum between conservative allocators and scale playersTed demonstrates deep sectoral expertise by contrasting Bank's deliberate decision to halt before agentic execution against Rowland's large-scale automated implementation at J.P. Morgan.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Ted as informed peer | Guest teaching | Guest disagreement | Ted pushing back | Why |
|---|---|---|---|---|---|---|
| Abby Barlow: AI as Sole Analyst at Westwood Management | 3 | 5 | 1 | 2 | Ted acts primarily as an inquisitive facilitator asking Barlow to detail her workflow with Claude as a solo analyst. He presses lightly on governance and manager confidentiality concerns regarding feeding external pitch decks into LLMs. | |
| Laura Hill: Preserving Human Judgment at Advocate Health | 3 | 6 | 2 | 1 | Hill lays down firm philosophical boundaries about protecting human judgment and forbidding AI from writing investment thesis memos. Ted guides the discussion cleanly without pushing back against her thesis. | |
| Brian Sugrue: Standardizing Workflows and Red Teaming at Shannon Bridge | 3 | 6 | 2 | 1 | Sugrue explains how red teaming with AI sharpens diligence meetings while debunking the myth that AI offers exponential time savings. Ted prompts on what failed and how decision-making remains human-dominated. | |
| Jenny Heller: Trust, Verification, and Operating Systems at Brandywine | 3 | 6 | 1 | 1 | Heller details multi-layered enterprise trust, verification discipline, and the failure of mid-diligence memo generation when context became too broad. Ted conducts structured discovery without challenging her operational framework. | |
| John Lawrence: Productivity and Operational Efficiency at Rice Management | 2 | 5 | 1 | 1 | Lawrence runs through specific productivity wins at Rice Management across summaries, models, and board presentations. Ted asks straightforward questions regarding portfolio integration. | |
| Sponsor Break: Ridgeline Modern Investment Management Platform | 2 | 4 | 0 | 0 | A mid-roll sponsor ad transition for Ridgeline followed by John Lawrence concluding his tech stack and alpha expectations with Ted playing a purely supportive host role. | |
| Matt Bank: Deconstructing Workflows and Internal Development at GEM | 3 | 6 | 2 | 1 | Bank outlines how GEM flipped from 90 percent vendor tools to 90 percent in-house development while deliberately holding back on full agentic decision-making. Ted asks about cost realities and operational constraints. | |
| Kristin Kallergis Rowland: Deploying AI Agents at J.P. Morgan | 3 | 7 | 1 | 1 | Rowland presents institutional-scale agent deployment across $250 billion in alternatives with 50 dedicated engineers. Ted asks specific tactical follow-ups on manager profiles and quantitative sentiment signals. | |
| Jon Webster: Institutional Memory and EQ Moats at CPPIB | 4 | 7 | 2 | 1 | Webster articulates CPPIB's philosophy of reading, remembering, and challenging everything, delivering a deep strategic point on why commoditized IQ shifts competitive advantage to organizational EQ. Ted steers the macro institutional narrative. |
Statements from this episode (29)
Barlow uses AI as her analyst instead of hiring one
“Part of the reason I have not hired is because I happen to start at the same time that AI was becoming a thing and have leaned into it to support me and become my analyst.”
Barlow says Microsoft Copilot currently trails Anthropic's Claude for investment work
“We also have a co-pilot subscription. It is not as good at this moment.”
Prompting AI to find its own errors catches missed mistakes, Barlow says
“I have gotten in a habit where I will ask it at the end of any deliverable, go back to the beginning of this conversation and what we were trying to accomplish, review the deliverable, triple check all the numbers again, find your own errors. It will be like, …”
Barlow says fund managers expect LPs to feed pitch decks into AI
“I have been asking managers, how do you feel about LPs taking your legal docs, pitch decks, and putting them into AI? The answers I've gotten have been all over the place. In general, they say, we just expect it's happening. We would assume that those LPs are …”
Allocators must not use AI for core risk questions, Hill says
“To me, the purpose of an investment memo is to sit down and make yourself think, what are the bets I'm taking? What would make this blow up? If we come back for a re-up in three or four years, what would make us pause or what could have gone wrong? I absolutel…”
Advocate Health uses internal AI to ingest private market statements, Hill says
“Right now, it's ingesting all of our private markets, statements, capital calls. It's extracting the data, which is different across every manager. It's fitting it out into a uniform Excel template that our analysts can then review.”
Hill is extremely skeptical of vendor AI document routing tools
“The external vendor solutions so far for document routing has not met expectations. We have chosen to build that internally. A lot of this is the virginous nature of how GPs send documents. I'm extremely skeptical of the solutions that claim to be able to mana…”
Allocator edge comes from relationships and decision speed, not AI, Hill says
“It's not a better PowerPoint. It's not a better model. It's not a better memo. AI can do all of those, but it's speed to decision, differentiated relationships, differentiated viewpoint.”
Shannon Bridge avoids AI for investment decisions or recommendations, Sugrue says
“The one thing we have not used it for is as a decision maker or a recommendation tool, because I'm not that comfortable with the idea of that.”
AI speed gains in diligence prep are modest, Sugrue says
“What I'm describing is a time-consuming preparation process to get to an answer. It's not two or three or four times faster than it would be otherwise. It's a little bit faster, and it facilitates the final steps, but all of the preparation on getting data qua…”
Brandywine will sandbox AI agents before deployment, Heller says
“If we ever use agents, which we're not doing yet, they'll be walled off and sandboxed before we let them loose.”
Brandywine's experiment using AI for mid-diligence updates failed, Heller says
“We tried to pull together mid diligence updates, and this would sit between our issue memo, which is early stage diligence sharing with the team, and our investment memo. The thought was that the lead or associate director could pull together this update using…”
LLMs lose accuracy with overly broad problem contexts, Heller says
“There's real limitations if the problem site gets too big and too unstructured. The LLMs start to lose their potency and their accuracy. The more structured and specific the question, the better the answer.”
Brandywine will not connect Claude to email until security is verified
“I would love to have Clyde be able to connect with my email. I'm just not at a place yet where we can be sure that that's safe and secure. Until we get there, we're approaching it slowly.”
Sam Lessin will teach NY investment teams to code with Claude
“We're going to have a Learn to Code through Cloud Day that one of our wonderful managers, Sam Lesson, is going to teach a number of teams in the New York area.”
Rice Management saves a day per board memo using AI, Lawrence says
“On the front end, on our board memos,
It's been tremendously positive beneficiary.
We save about a day of time.
That's a very onerous process.
It consolidates the work that our team has already done and puts it into a template for us.”
Claude is effective at building financial models from scratch, Lawrence says
“Building valuation models and building financial models.
Claude is good at this, and it's been helpful when we start from scratch.”
Canoe saves Rice Management operations roughly 30 hours, Lawrence says
“From operations perspective, processing incoming documents, the tool canoe, that's been helpful.
Automating a lot of the integration of those documents into our current systems saves roughly, 30 hours of time for our operations team.”
Rice Management's centralized data lake house is delayed, Lawrence says
“The real challenge is aggregating our data into one centralized data lake house, which we're working on. It's easier said than done. That has taken more time than I would have expected. I was hoping we would have had this six months ago. We're still working on…”
GEM shifted to 90 percent internal software development in one year
“However, if you rewind the calendar a year, about 90% of our spend and our tool set was externally developed. We were buying off the shelf things, trying to integrate them. Today, that has flipped. 90% of the things we're developing are internal.”
GEM outputs 30,000 lines of code monthly, up tenfold, Bank says
“Our team generates about 30,000 lines of code per month. I'm told that's a lot. I don't know for a firm of our size, but it's up 10 X from where that was last year.”
Agentic AI is not yet cost-effective for workflows, Bank says
“There's agentic chatter about the capacity of the tool to take workflows end to end and execute on them. We haven't really seen that in a cost effective way work. We've had to meter some of our token usage because of things that were way out of bounds with res…”
AI has not cut GEM headcount and is no panacea, Bank says
“We still have all the same people. In fact, we've added folks. We will continue to add folks who will be capable because of this tool, but it is by no means a panacea for operational expense.”
AI reduced J.P. Morgan alternatives workload by one third, Rowland says
“It's taken off probably a third of the workload of people just in the last six to nine months.”
Over 50 AI engineers manage J.P. Morgan alternative assets, Rowland says
“I have over 50 engineers that work on two hundred fifty billion dollars of assets that we oversee. Everyone's an AI engineer.”
AI reduced J.P. Morgan tech mockup time to weeks, Rowland says
“Every time we go to create what our tech agenda is for where we want to deploy AI next, it would take us between two and a half to three months to implement that idea. Now that we have AI embedded in every aspect of what we do, it takes us two to three weeks t…”
LLMs enable CPP Investments to cover a wider filings universe, Webster says
“Natural language processing with LM is allowing us to read filings and earning calls transcripts across a much broader universe than any team could have historically covered.”