Summary
Pressure to achieve AI outcomes reaches every part of investment management, but the runway differs by segment. Three variables govern the pace: in-house build capacity, a mature system of record, and the room budgets and vendor contracts allow. Each firm should find the variables that matter.
Pressure to show AI results now reaches every segment of investment management. Clients ask their vendors for it, and prospects choose vendors based on it.
Each segment’s runway depends on three things: how much a firm can build in-house, whether it has a system of record to put AI on, and how much its IT budget and vendor relationships constrain it.
These constraints resolve differently across hedge funds, private markets, institutional asset managers, and the sell side, which is why a segment-by-segment view tells you more than a single read on adoption.
Hedge funds pair small teams and flat hierarchies with a drive to act on information faster than rivals. That speed carries into technology. They adopt quickly and have the engineering depth to build agentic AI directly into investment workflows, which puts them furthest along the runway.
In front-office workflows, AI agents can accelerate research, synthesize information, generate investment ideas, support manager decisions, and build directly on cloud infrastructure. Middle- and back-office functions present different circumstances. They run on a patchwork of systems from multiple vendors, covering accounting, reconciliation, reporting, compliance, and operations. Each vendor in those areas is maturing at a different pace.
According to OpenAI’s case study with Balyasny Asset Management,i 95% of the firm’s roughly 180 investment teams actively use an AI-native research platform powered by frontier models. Document-intensive workflows that once required days to process are compressed into a matter of hours, reducing macroeconomic scenario analyses from two days to roughly 30 minutes.
How far a hedge fund can go depends on its build capacity and budget. Firms with large IT budgets can put dedicated teams on operational AI agents. Firms with tighter budgets have to weigh the cost of building against upgrading the accounting system, usually choosing one or the other. That pushes the pressure onto vendors. Hedge funds now expect integrated, production-ready AI from their providers, and a provider that cannot deliver gets bypassed by clients that build around it or move to a more modern, API-first alternative.
For decades, the private markets segment has run on bespoke point solutions, specialized software for deal analysis, valuation modeling, portfolio monitoring, and due diligence. General-purpose AI platforms like Claude for Financial Servicesii now perform much of that work faster and at a fraction of the cost.
However, there’s a limit to general-purpose platforms. Managers often have years of deal and investment history, decision frameworks, portfolio company information, and operating metrics in disconnected Excel spreadsheets, emails, and manually maintained processes. Agents can’t be deployed on a sea of spreadsheets because LLMs are built to deliver probabilistic outputs. When you have fragmented, non-standardized data inputs, you end up with inconsistent, unreliable results that no regulated workflow can rely on.
For that reason, a mature system of record remains an essential first step in private markets. You need to capture the operational and environmental context before agents can function reliably. Standardized data and workflows are prerequisites for automation. Without them, AI use in private markets will likely remain confined to narrow tasks like document summarization.
In addition, there’s an important caveat as point solutions and the vendors that developed them are squeezed out by frontier models. Managers who depend on widely available tools risk exposure of proprietary investment data, over-dependence on external AI providers, operational halts when a platform goes down, or unauthorized model training. Risks like these can directly compromise their performance or relationships with investors.
Institutional asset managers are adopting AI across the whole stack, from trading and execution to back-office operations and performance. Their constraint runs in the opposite direction to private markets. Vendor lock-in and tight IT budgets leave them with the fewest degrees of freedom, forcing a choice between augmenting with a slow incumbent and committing to a full modernization.
Many large asset managers also have deep, decades-long relationships with platform vendors that go beyond their investment management software. These enterprise platforms become embedded across operations, trading, risk, and compliance functions, and they’re hard to get rid of because of the operational switching costs. Scale, regulatory requirements, and deeply embedded client integrations make rapid rollout difficult, leaving them to evolve at a pace that lags technological potential.
With that infrastructure lock-in, many asset managers try to meet internal AI mandates with the tools their provider already ships. Such tools help summarize communications and draft internal materials but can’t stand in yet for agents that run portfolio construction, optimize trading workflows, or handle complex compliance work. A second AI platform on top of an existing contract is rarely in the budget.
By contrast, firms gain greater freedom to deploy advanced agentic capabilities across the organization when they pursue a modernization path to replace legacy systems wholesale, move to cloud-native architectures, and adopt AI-first tools from the ground up. This path also pressures incumbent vendors to accelerate their own AI roadmaps, giving firms more leverage to demand better tooling rather than having to build around or replace them.
The sell side splits into two. At the top of the market, Tier 1 global banks and prime brokers are pulling away from the rest of the industry by building almost all of their AI in-house. They’re investing billions in proprietary model pipelines, agent frameworks, and engineering talent on the bet that autonomous AI will lower their cost basis and create a durable advantage. Many already run production agents in reconciliation, exception management, and settlement breaks, where a compressed T+1 clock rewards speed and precision.
“The share of generative AI continues to grow as a percentage of our total AI activity. And overall, we’ve doubled the number of use cases in production this year. We’re focusing our efforts on the highest-impact areas such as customer service, including call center efficiency and personalized client insights as well as in technology, particularly, for our software engineers.” — Jeremy Barnum, Chief Financial Officer, J.P. Morganiii
These banks also use AI to reallocate junior talent toward oversight, validation, and exception handling.
Below Tier 1, the picture changes. Long technology development cycles, limited budgets, scarce technical talent, and regulatory constraints slow every rollout and narrow what these firms can attempt. They run into the same constraints the buy side does.
Even modernization does not address the segment’s most significant constraint — data readiness. It’s the same system-of-record problem private market firms face, spread across the entire structure of the business. Decades of mergers, siloed business lines, and layered infrastructure have left many banks defining a customer, an asset class, or a trading position differently from one system to the next. When the same entity carries conflicting definitions, the agents built on top inherit the conflict and return unreliable or contradictory output. The direction is the same across the sell side. The pace differs, however, and a bank’s data history is what sets it.
Across every segment, the runway comes down to the same three variables: what a firm can build, whether it has a system of record to put AI on, and the room its budget and vendor contracts leave. They resolve differently in each case. The work now is to find the variable that gates you and decide what it would take to move it this year. By the time the competitive advantage of agentic AI becomes clear, the cost of catching up will be prohibitive, and the runway will have already closed.
Dmitry (Mitya) Miller
Dmitry (Mitya) Miller is the Managing Director, General Manager for Aquata, Arcesium’s comprehensive self-service data platform purpose built for the investment management industry. Mitya is responsible for overseeing all aspects of the Aquata business, including P&L ownership, customer base growth, customer delivery and engagement, and product roadmap.
Sources:
i Open AI, “How Balyasny Asset Management built an AI research engine” March 6, 2026. https://openai.com/index/balyasny-asset-management/
ii Anthropic, “Claude for Financial Services,” July 15, 2025. https://www.anthropic.com/news/claude-for-financial-services
iii JPMorgan, 2026 Company Update, transcript, February 23, 2026. https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/investor-relations/documents/2026-company-updates/company-update-full-event-transcript.pdf
No spam. Just the latest releases and tips, interesting articles, and exclusive interviews in your inbox every week.
