ADOPTION
Citi: AI Evolving From Cutting Costs to Picking Investments
Citi has new research out this month on how AI is moving beyond efficiency gains in financial services to steering investment decisions.
The latest report from Citi’s data and markets research team builds on its June 2024 study of more than 40 investment executives. It updates those interviews with new conversations and input from Citi and Microsoft’s AI teams.
For those of you who read AI Street regularly, some of these use cases will sound familiar: summarization, transcription, sentiment analysis. Here’s a breakdown:
For Fundamental Investors:
Processing 30+ analyst reports simultaneously
Behavioral analysis of management communications for alpha generation
Forensic accounting and anomaly detection
Reduction in research synthesis time from weeks to hours
For Quantitative Investors:
Autonomous factor research and signal generation
Real-time portfolio monitoring and risk management
Time series forecasting with dynamic model selection
Investors Use Hallucinations to Challenge Investment Thesis
One of the report’s surprising use cases is investors using “hallucinations,” where AI makes up info, as an advantage. Some firms are using errors as brainstorming tools. Think of it this way: When an AI hallucinates, it's essentially making unexpected connections or suggesting scenarios that might not be grounded in its training data.
Increasingly, investors are leveraging hallucinations positively to explore additional considerations and dimensions not previously thought about, transforming a limitation into an opportunity for deeper and more creative analysis.
AI Could Make Bonds Trade Like Stocks
Citi says Large Language Models are starting to close the automation gap between equities and fixed income.
About 40% of credit markets trade electronically today. The rest is still manual, slowed by complex, bespoke documentation. LLMs can read and extract key terms from that unstructured text—bond covenants, collateral clauses, swap conditions—making it easier to price and trade debt products automatically.
The equity markets are more structured than the bond world, so AI has more room to make an impact there.
I wrote back in February:
The line between private and public markets is blurring because AI lets investors standardize information buried in PDFs. Public companies have regulatory requirements in how they report their financials. In the private world, there’s often less information and it’s not necessarily presented in an easily digestible way.
Emerging AI Use Cases
Citi flags agentic AI —where the tech completes user specified tasks autonomously — as the next leap. I also came across some topics that I’m not too familiar with:
Data Mesh Architecture — managing data without one big central warehouse by letting each team, like sales and marketing, own and maintain its own “data product” while still sharing across the company.
Graph-based signal discovery — using network maps to reveal relationships among companies, sectors, or factors that traditional models might miss.
Takeaway
Citi calls generative AI a seismic shift for both fundamental and quant investors. The technology isn’t all the way there yet, and scaling will take years, but this is a good roadmap for where the Street is headed.
Further Reading
AI in Investment Management - Beyond Efficiency Gains - Citi
FUNDRAISING & PE
Menlo Ventures Backs AI Startup For PE
The venture firm led a $7.1 million seed round for Trove AI
Menlo Ventures led a $7.1 million seed round for Trove AI that builds AI agents to handle routine work at PE firms.
Trove’s software helps investors automate deal screening and portfolio reporting. The system runs in a firm’s private cloud and builds a knowledge graph of internal data, allowing it to retrieve and analyze information quickly and securely.
Early customers include Shamrock Capital, which has $6.6 billion in AUM and NewView Capital with $3 billion in AUM, per Trove.
Much of the attention around AI in finance has centered on equities. But private equity firms have a real need for AI data extraction tools, given how scattered their data sources are.
I asked CEO Danny Goldman about how they mitigate hallucinations. He said that RAG, which narrows the number of documents an LLM reviews to the most relevant ones, doesn’t work well when capturing relationships across companies, deals, docs, people, etc. So, Trove builds a knowledge graph on top of a firm’s internal data. (A topic I wrote about last week: Intro to Knowledge Graphs.)
Goldman said the clearest demand so far is in deal sourcing:
Initial deal screening is the most popular use case right now - high frequency, clear ROI, and our agents can do much more work than a normal associate can in a much quicker timeframe (often tackles work that wouldn't happen otherwise).
This quote reminded me of an interview I did last year with Bain’s Richard Lichtenstein on AI adoption in PE. He noted that sourcing can be tedious: “You might look at 20 deals before you find one company that you get excited enough about. That's not an amazing use of people's time… AI can really help with that.”
Takeaway
AI in finance has mostly centered on equities. But private equity data is scattered across sources, formats, and systems. AI tools can help cut down on analyst grunt work.
Further Reading
M&A
OpenAI Buys Personal Finance App Roi
Made with ChatGPT
OpenAI makes headlines almost daily, but this acquisition caught my eye:
Roi, an AI-powered personal finance app, aggregates a user’s financial footprint, including stocks, crypto, DeFi, real-estate, and NFTs, into one platform to track funds, analyze performance, and execute trades, per TechCrunch.
Roi’s portfolio breakdowns mirrored a user’s tone, even roasting—meaning teasing (for anyone over 50 🙂)—them over losses. This is a far cry from today’s investor interfaces but points to where things are headed: personalized, conversational experiences built on LLMs.
Takeaway
Large language models can customize your Spotify playlist, your news feed, and your investment dashboards -- at scale. In finance, expect personalized portfolios tuned to your tone and style.
Further Reading
Q&A with Arta CIO on AI-Driven Wealth Management - AI Street
OpenAI’s Strategy Echoes Microsoft’s Windows Playbook - Stratechery
NEWS
AI’s Muted Impact on Bank Jobs
Banks posted strong earnings this week and offered a few updates on how AI is showing up inside their operations.
Citi said its AI tools are now saving around 100,000 developer hours each week.
Goldman Sachs mentioned that AI is part of an operational overhaul—known internally as “OneGS 3.0”—which includes a limited reduction in roles and a freeze on headcount growth.
Recent reporting suggests that AI may let JPMorgan hire fewer people over time, particularly in support and operations roles.
So far, I haven’t seen any meaningful evidence that AI is causing mass layoffs in major sectors. That aligns with what the Fed has said: AI will lead to some job losses, but is more likely to lead to retraining.
Takeaway
Much of the coverage around AI assumes the technology is inevitably coming for everyone’s job. For now, there’s little data to support that view.
Further Reading
SPONSORSHIPS
Reach Wall Street’s AI Decision-Makers
Advertise on AI Street to reach a highly engaged audience of decision-makers at firms including JPMorgan, Citadel, BlackRock, Skadden, McKinsey, and more. Sponsorships are reserved for companies in AI, markets, and finance. Email me ([email protected]) for more details.
ROUNDUP
What Else I’m Reading
AI’s Growth Leaves Financial Regulators Struggling to Catch Up - BBG
Ex-Citi Execs launch AI platform for family offices - P&I
Griffin Says AI Fails to Help Hedge Funds Beat Markets - BBG
AI Can Transform the Economy and Still Be a Bubble - BBG Opinion
Senate Backs Treasury Study on AI Anti-Money Laundering - FedScoop
Lloyds’ Execs to Go Through Six-Month AI course - MSN
Walmart Partners With OpenAI to Offer Shopping on ChatGPT - BBG
Salesforce Says AI Customer Service Saves $100 Million Annually - BBG
LSEG and Microsoft to Help Banks Build AI Agents - PYMNTS
How WSJ Readers Think AI Will Change Wall Street - WSJ
California Passes First State Laws Regulating AI Chatbots - CNBC
CALENDAR
Upcoming AI + Finance Conferences
*Recently added events in bold
Open Source in Finance Forum - Oct. 21-22 • New York
Finance and tech leaders tackle how open source and AI can be governed, scaled, and applied in financial services.
AIFin Workshop at ECAI 2025 – October 26 • Bologna, Italy
One-day academic workshop on AI/ML in finance, covering trading, risk, fraud, NLP, and regulation.
AI in Finance 2025 – October 27–30 • Montréal
Academic event covering ML in empirical asset pricing and risk.
Women in AI NY x ACM-W NYC - Nov. 7 • New York
AI scientists, policy experts, and senior data leaders discuss bridging theoretical research with practical applications.
ACM ICAIF 2025 – November 15–18 • Singapore
Top-tier academic/industry conference on AI in finance and trading.
AI for Finance – November 24–26 • Paris
Artefact’s AI for Finance summit, focused on generative AI, future of finance, digital sovereignty, and regulation
NeurIPS Workshop: Generative AI in Finance – Dec. 6/7 • San Diego One-day academic workshop at NeurIPS focused on generative AI applications in finance, organized by ML researchers.