AI and Data

Citadel, Balyasny push proprietary AI deeper into analyst and PM workflows

Ken Griffin says work that took PhDs weeks now runs in days, shifting the pressure onto portfolio managers to convert output into P&L.

INFLXD Research·August 24, 2026·3 min read

Citadel, Balyasny push proprietary AI deeper into analyst and PM workflows

Hedge funds are moving AI tooling out of the research back office and into the seat of the analyst and portfolio manager, according to a feature published by eFinancialCareers on Monday. Citadel and Balyasny are the two names carrying most of the weight in the piece, and both have platforms already sitting inside the daily workflow of their investment staff.

Citadel rolled out an in-house AI tool for its equities traders last year. Then-CTO Umesh Subramanian said in December the tool flags risk and generates reading lists for traders based on their portfolios, per eFinancialCareers. Ken Griffin said in May that complex research that would take finance PhDs weeks can now be done by AI agents in days, according to the same report.

At Balyasny, OpenAI disclosed in March that 95% of investment teams were using the fund's proprietary AI platform, which includes a deep research tool designed to replicate the work of a senior investment analyst in a fraction of the time, per eFinancialCareers citing OpenAI. The tools ingest complex documents and score their potential portfolio impact.

An oversized hourglass where the upper chamber is filled with dense transcript pages and marked-up analyst notes draining rapidly into the lower chamber , which is a Bloomberg-style terminal panel alr

The framing in the eFinancialCareers piece is that speed of research is no longer the constraint. If a deep-research agent can compress weeks into days, the bottleneck moves to the portfolio manager, who has to translate that output into positions and sizing. The article makes that point directly: faster insight generation places more pressure on PMs to convert insight to profit.

The hiring signal is the other half of the story. eFinancialCareers argues the skills that matter for analysts now overlap heavily with what matters for PMs: novel use of AI tools, and the ability to identify alternative data streams that improve a trading strategy, even when the grunt work is automated. In other words, the differentiator is no longer speed of processing. It is judgment on what to process and what to do with the answer.

Why it matters

Two things worth watching from here.

First, the deep-research agent claim. Balyasny's tool, and Griffin's "weeks to days" line, both describe systems that ingest large document sets and produce analyst-grade output. That is the right comparison for the grunt work: reading filings, pulling comps, synthesising transcripts. It is a harder claim to make for the parts of the analyst job that involve calling an expert, weighing management credibility, or deciding what a competitor is not saying. Our read is that the productivity gains here are real for information retrieval and initial synthesis, and much less established for judgment-shaped work. The buy-side will find out which is which over the next few earnings cycles.

Second, the pressure shift onto PMs is the more interesting structural point. If two multi-manager platforms are running proprietary agents across nearly all of their investment teams, the marginal advantage of "we did more research" collapses inside those firms. What is left is sizing, timing, and risk. That is a harder game to win, and it puts a premium on PMs who can act on faster information without over-trading it. Expect the compensation and retention conversation at multi-managers to move accordingly.

For expert networks

Are your experts using AI to cheat?

Try our free demo to find out today.

We help you detect the use of AI on expert calls.

Try demo →INFLXD.

What to watch: whether other multi-manager platforms disclose comparable adoption numbers, and whether the next round of analyst job specs at Citadel, Balyasny, Millennium and Point72 make AI-tool fluency an explicit requirement rather than a nice-to-have.

From INFLXD

Powering institutional-grade transcription for expert networks.

INFLXD provides AI-powered, human-edited transcription with sub-1% error rates for the world's leading expert networks and financial research firms.

Visit inflxd.com →

Trending now

Keep reading.

Markets

Buy-side research budgets flat in H1 2026 as AI spend climbs 40%

Overall research wallets grew just 0.4% in the first half, with the reallocation flowing into AI tooling rather than external providers.

INFLXD Research · Oct 9

M&A

BlueMatrix to acquire Aiera, backed by Thoma Bravo

The combined platform targets governed distribution of broker research into buy-side AI workflows.

INFLXD Research · Oct 8

Markets

ABI Research names Starlink and Skylo leaders in D2C satellite ranking, Elveo flagged on innovation

The direct-to-cell satellite competitive map now has a public scorecard, with Elveo Mobile positioned as the challenger behind Lynk and Omnispace's merger.

INFLXD Research · Oct 8