Hey, I’m Matt. I’m a former Bloomberg News reporter, and you’re reading AI Street, where I report on how Wall Street uses AI. Every Thursday, I cover the week’s biggest AI-in-finance stories, plus original reporting and analysis.
Lots of news this week, so this edition is more of the quick-hit variety.
Bridgewater Says Controlling AI Agents Remains an ‘Unsolved Problem’
Bridgewater, Citadel and Man Group are expanding their use of AI, but risks limit how much they let agents do on their own, according to Business Insider.
Bridgewater partner Aaron Linsky said ensuring agents don’t put the firm at risk remains an “unsolved problem.” Citadel’s interim technology chief, Andrew Janian, said the firm spends a lot of time making sure agents stay within their limits.
Man Group investment chief Greg Bond warned against “cognitive surrender,” where people hand over their work to AI and go on autopilot.
PRESENTED BY COMPOUND
Compound leads the efficient frontier on real private equity work
Can AI do the work of a PE analyst? We built a benchmark to find out.
Introducing PE Bench, the first benchmark for AI agents on real private equity work. Across 5 data rooms, agents build diligence cuts, an operating model, and an IC memo and deck.
Compound sets the efficient frontier against agents from every leading lab: our standard mode ranked first, ahead of Claude Code on Opus 5.5 at two-thirds of the cost, and our quick mode beat Codex on GPT-6 Astra at half the cost.
These results come from Compound V2, which we launched two weeks ago: your autonomous AI deal team. A VP orchestrator runs specialized agents across every deal, your colleagues share the same agents in Compound or over email, and a proactive agent tracks filings, earnings and news on the companies you follow.
See the new Compound in action:
The Arms Race on Wall Street
Bloomberg had a couple of stories out this week on topics we’ve been talking about: the arms race for GPUs, how it parallels the rise of HFT and co-location in the 2010s, and the demand for talent.
Jane Street and Rivals Are Now Fighting for AI Power as Well as Pace
AI Talent War Has Quant Firms Racing to Fund PhDs, Build Labs
AI Tracker
I’ve created a database that follows what financial firms have publicly said about how they’re using AI. It includes 86 entries covering 56 firms, including Bridgewater, Two Sigma, Man Group, AQR and Balyasny.
The entries draw on interviews, conference talks and technical posts, with links to the original sources and key takeaways. You can search by firm, person, use case or keyword.
Paid subscribers can access the tracker below:
Itaú’s $580 Million Bet on Foundation Models
Itaú, the Brazilian bank, agreed to buy a startup that builds AI models for about $580 million. The startup, NeoSpace, had already trained models on the bank’s own data.
According to an Nvidia case study, NeoSpace uses transformers, the architecture behind modern language models. The idea is to build one foundation model that can be adapted to different tasks, rather than relying on a bunch of separate models and processes that banks typically use.
I’ve been fascinated with the idea that if you have enough computing power and enough data, you can train a model that gives you pretty good results. (In fact, models trained on weather data can outperform those based on atmospheric physics.)
This is a topic I’ve covered a few times here:
“Based on all the conversations that we’ve had with clients, everybody’s preparing for this,” said Zornitsa Todorova, who leads thematic fixed-income currency and commodities research at Barclays, referring to AI bots executing more trades. “It’s a slow burner, but it’s definitely going to happen.”
From AI Agents Seen Taking Credit-Trading Work From Humans Over Time
Few Financial Firms Give AI Agents Broad Autonomy: Survey
There's no industry group that tracks AI in finance (at least for now) like the Mortgage Bankers Association does for mortgages or the Investment Company Institute does for mutual funds. So there’s not a ton of data beyond vendor surveys.
But the FCA and Bank of England have surveyed financial firms about their AI use every other year since 2022, and the latest one came out this week.
Agents are widespread, but mostly assist humans: 80% had deployed agents, but only 2% reported broad autonomy.
Agent autonomy isn’t expected in high-stakes transactions. No more than 9% expected agents to execute trades, handle payments, approve or decline customer outcomes, or change controls.
AI adoption is uneven: Ten firms accounted for 43% of all reported AI systems, while almost half of respondents had ten or fewer.
The regulators surveyed 196 financial firms operating in the UK, including banks, insurers and asset managers. The agent findings are based on the 105 firms that completed the full questionnaire.
The full report is lengthy and worth checking out.
Morgan Stanley’s Agents Help Staff Move Money
Other financial firms may be doing something similar, but this is the first time I’ve heard of an institution using agents (with human supervision) to help move money. “The agent can retrieve information and take actions such as money movement, opening accounts and changing of beneficiaries,” Banking Dive reports.
Anthropic Adds More Tools for Financial Advisers
Anthropic has added more partners to Claude for Financial Advisors, including Invesco, FINNY, CAIS, Conquest and J.P. Morgan Asset Management’s 55ip, according to a company announcement.
We covered the launch in September. This adds to Anthropic’s effort to make Claude the place where advisers do more of their day-to-day work.
AQR Partners Invest in AI Startup LinqAlpha
Some partners at AQR have personally invested in AI research startup LinqAlpha, according to a company statement. The startup didn’t name the partners or disclose the size or timing of their investment.
AI Is Not Taking Your Job
As I’ve said many times before, there’s no clear evidence that AI is leading to mass unemployment. In fact, new technologies have historically created more jobs than they’ve eliminated. Yes, jobs change, and some are eliminated, but new ones pop up. From CNBC:
U.S. software development job postings have risen since February 2025, the launch date for Anthropic’s Claude Code, even as overall job postings fell, complicating predictions that AI would rapidly erase demand for developers.
Another profession that seems like AI should replace: writing. But Anthropic is looking to hire an “Executive Communications Writer” at a salary that tops out at $345,000. No one wants to read ‘it’s not X, it’s Y’ for the millionth time.
ROUNDUP
What Else I’m Reading
Wall Street regulator to focus on AI, private assets in corporate exams | Reuters
FDIC’s Hill mulls agentic AI deposit impact | Banking Dive
Two Sigma Said to Explore Stake Sale to End Founders’ Stalemate | BBG
Bloomberg Adds AI Coding Assistant to BQuant | PR
Cognitive Credit Raises $40M for Credit Data and AI | Finextra
Corporate-Bond Trading Is Moving Beyond Top Securities, Barclays Says | BBG
Smaller AI models help BofA, Capital One manage costs | CIO Dive
Sponsor AI Street
November sponsorships are SOLD OUT. Limited December placements are available starting Dec. 3.
Reach more than 5,000 institutional investors and decision makers across finance, AI, and law. To discuss available dates and sponsorship fit, email sponsors@ai-street.co or fill out this form here.
This Week in AI Street
Model ML Seeks Funding at $1.5 Billion Valuation
Since emerging from stealth in February 2025, Model ML has:
-- Named HSBC, FT Partners, PwC and Deloitte among its clients.
-- Added advisers including former HSBC CEO Noel Quinn, former UBS chairman Axel Weber and former Julius Baer CEO Philipp Rickenbacher.
-- Reported doubling annual recurring revenue and adding more than 47,000 seats in the second quarter of 2026, (w/o disclosing total revenue.)
-- Grown to nearly 200 employees across nine offices.
-- Raised more than $100 million, including a $12 million seed round and a $75 million Series A.
Model ML is now seeking another funding round that could value the company at about $1.5 billion, according to a person familiar with the matter. Bloomberg reported in September that the company was discussing a raise of more than $100 million, potentially reaching $150 million, at a valuation above $1 billion. A $1.5 billion valuation would put Model ML below rival Rogo, valued at $2 billion, but above Hebbia, whose last reported valuation was roughly $700 million.
A Model ML spokesperson didn’t respond to requests for comment. Chaz Englander also didn’t respond to a LinkedIn message seeking comment for this story.
CALENDAR
Upcoming AI + Finance Conferences
Recently added events in BOLD.
Private Markets AI Summit — November 4 • New York
Private Markets Forum summit on AI agents in deal screening, investment operations and portfolio monitoring, with participants from Apollo, BlackRock, Ares and CPP Investments.Open Source in Finance Forum New York — November 4–5 • New York
FINOS conference for financial-services technologists featuring keynotes, panels and technical sessions on open-source infrastructure and collaboration.FinTech Founder & Investor Evening — November 4 • Frankfurt
Invitation-only evening hosted by Finteda, Schalast and Fincite, bringing together founders and investors to discuss capital, regulation and scaling fintech companies in Europe.Columbia MAFN & AI Conference — November 5 • New York
Rebellion Research-hosted Columbia conference on AI and machine learning in portfolio management, with speakers from AQR, Two Sigma, and Citadel.Asset & Wealth Management Summit — November 10–11 • Boston, MA
Nicsa summit on alts and AI transforming asset/wealth management, with speakers from Anthropic, OpenAI, Morgan Stanley, KKR and Microsoft.21st SFI Annual Meeting — November 12 • Zurich
Swiss Finance Institute’s annual meeting on AI’s impact on finance, with keynotes from UBS, ETH AI Center and Oxford on AI’s risks and opportunities in banking and trading.ICAIF ‘26 — November 14–17 • Milan
7th ACM International Conference on AI in Finance, bringing together researchers and practitioners on generative AI, trading, risk, fraud detection and responsible AI, hosted at Bocconi University.AI in Financial Services Conference USA — November 18 • New York
GlobalData conference on generative and agentic AI deployment across banking, insurance and wealth management.
Thanks for reading!
I’m always happy to receive comments, questions, and feedback.
Connect with me on LinkedIn, or
Send an email to matt [at] ai-street.co
Manage how often you receive AI Street
AI Street is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.