When people picture “investment research,” they often imagine giant spreadsheets and long PDF reports.
Those tools still exist. But today a growing share of the work also runs through AI models that read text, scan data, and surface patterns long before a human opens Excel.
This article explains how analysts on the buy side (asset managers, pension funds) and sell side (brokerage and bank research teams) use AI behind the scenes. It stays descriptive and neutral. It is not investment advice.
For a look at how similar tools power automated portfolios for end clients, see:
How AI Robo-Advisors Build Portfolios.
1. What “investment research” means in practice
In a typical firm, “investment research” covers several jobs:
- Understanding companies, sectors, and themes
- Reading financial statements, filings, and earnings call transcripts
- Tracking news, macro data, and policy decisions
- Building models to estimate risks and potential return scenarios
The IMF and IOSCO both describe AI in capital markets as a set of tools that now touch data gathering, signal extraction, portfolio analysis, and risk management across buy-side and sell-side firms.
So AI is not one product.
It is a layer that shows up at almost every step of this workflow.
2. How widely is AI actually used today?
A recent CFA Institute handbook on AI in asset management and a 2025 workflow report on big data note that adoption has moved from “experiments” to more routine use, especially in research and data tasks.
At the same time, a CFA Institute survey in 2024 found that:
- Most investment employers want standards and training for AI use.
- Many firms feel unprepared, even as employees expect to use AI more.
Another CFA article in 2025 points out something important: only a tiny fraction of funds officially describe AI as part of their stated strategy, but AI tools are already common in the backstage research process.
In short: AI often supports human research quietly, even when the fund’s label never mentions it.
3. Data first: from classic numbers to “alternative data”
Good research starts with data. AI changes what data analysts can use and how they handle it.
Traditional market data
This includes:
- Prices and volumes
- Index levels
- Company financials and ratios
AI helps by:
- Cleaning and organizing large time series
- Detecting outliers or data-quality issues more quickly
- Linking related securities or issuers across regions and asset classes
Alternative data
Reports from the IMF and IOSCO describe how buy-side firms now explore alternative data with AI, such as:
- Web traffic, app downloads, and search trends
- Satellite images or shipping data
- Anonymized credit card or transaction data (where allowed by law)
Machine learning models can turn this messy information into:
- Indicators of business activity
- Early signs of changing consumer behavior
- Signals that may not appear in official financial statements yet
CFA Institute’s big data workflow study shows that many teams now treat AI as a bridge between raw data sources and the traditional research process.
4. Natural language tools for filings, transcripts, and news
One of the biggest changes comes from language models that can read text at scale.
Regulatory filings and reports
Analysts have always read:
- 10-K and 10-Q filings
- Annual reports and presentations
- Prospectuses and risk disclosures
Now, large language models can:
- Extract key risks or themes from long documents
- Compare wording across years to spot subtle changes
- Flag unusual phrases or disclosures that deserve human follow-up
The IOSCO report on AI in capital markets notes that natural-language tools are widely used to scan regulatory filings and research reports and to create forward-looking indicators from text.
Earnings calls and corporate communication
LLMs also summarize:
- Earnings call transcripts
- Investor day presentations
- Management Q&A sessions
BlackRock and other large managers describe using machine learning on speech and text to capture tone, themes, and sentiment across many companies at once.
This does not replace the analyst.
Instead, AI:
- Surfaces unusual comments faster
- Helps compare language across competitors
- Saves time on note-taking so humans can focus on interpretation
For a user, much of this shows up as:
- “Key takeaways” sections
- Faster post-earnings notes
- Dashboard alerts when something important changes in language
5. Pattern detection and “signals” in markets
AI also helps find patterns that are hard to see with traditional tools.
Macro and cross-asset signals
BlackRock’s research on machine learning in macro investing explains how purpose-built models can turn scattered signals—such as multiple economic indicators, asset prices, and policy variables—into global views of risk and relative value.
The IMF’s 2024 chapter on AI and capital markets reports that buy-side firms use AI to:
- Explore new asset classes
- Extract trading signals from complex datasets
- Support portfolio allocation and scenario analysis
Equity and credit research
CFA Institute’s AI monograph and earlier “AI Pioneers” report list use cases such as:
- Ranking stocks or bonds by multi-factor models enhanced with machine learning
- Detecting unusual combinations of metrics that might signal stress or strength
- Testing many variations of factors more quickly than classic tools allow
Other articles describe internal AI labs at large firms that experiment with systematic equity models combining fundamentals and AI-driven features.
The key point: AI focuses on patterns and probabilities, while humans still interpret whether those patterns make economic sense.
6. Risk and scenario analysis with AI platforms
AI also plays a growing role in risk systems.
BlackRock’s public material and third-party commentary describe its Aladdin platform as a long-running system that uses AI and machine learning for risk analytics, scenario testing, and portfolio monitoring across trillions of dollars in assets.
IOSCO and IMF documents mention similar uses at other firms:
- Stress tests under many simulated scenarios
- Early-warning indicators for liquidity or concentration risk
- Tools that combine positions, market data, and models to monitor exposures
Recent articles on Goldman Sachs and other banks highlight AI-driven models for risk assessment and anomaly detection, as well as trade allocation and pricing support.
These systems do not remove the risk of loss.
They try to measure and visualize risk more quickly and from more angles, which changes how analysts and risk teams spend their time.
For background on risk concepts themselves, you can link readers to:
Investment Risk and Market Volatility Explained.
7. AI as a productivity tool: research assistants and portals
A big part of the story in 2025 is simple: saving time.
Recent news shows several examples:
- Citadel launched an AI Assistant that helps equity investors search transcripts, filings, and research, then compile tailored reports relevant to their portfolios. The CTO described it as a way to reduce research time while keeping humans in charge of decisions.
- BNP Paribas rolled out an AI-powered portal to cut pitch-preparation time, mining previous materials and data so bankers can focus more on strategy.
- Other banks, including JPMorgan, Goldman Sachs, UBS, and Nomura, have announced internal AI copilots or LLM suites to assist with analysis, drafting, and presentations.
CFA Institute’s 2025 workflow report similarly notes that investment professionals are using AI and big-data tools to complement traditional software like Excel, with strong interest in upskilling across the industry.
For the analyst, this often means:
- Fewer hours on manual data pulls and formatting
- More time on high-level judgment and talking with companies
- Faster ability to test “what if?” questions and scenario tweaks
Again, the decision remains human.
AI shifts where the effort goes.
8. Limits, risks, and the push for standards
Regulators and professional bodies also talk about what AI cannot do safely on its own.
Model risk and “black boxes”
IOSCO, the IMF, and CFA Institute all highlight model risk:
- Complex AI systems can be hard to interpret.
- Bad data or biased training sets can produce misleading signals.
- Overfitting to past data may fail when conditions change.
The TABB Forum “State of AI in the Capital Markets 2025” report adds that data fragmentation and hallucinations remain practical obstacles, even as adoption speeds up.
Governance and ethics
CFA Institute’s 2024 survey found broad agreement among industry professionals that:
- The industry needs clearer standards for AI use.
- Firms must invest in governance, transparency, and workforce training.
- Many people feel both optimistic and anxious about AI’s impact on their roles.
Regulators also stress that using AI does not remove a firm’s responsibility to:
- Know its clients and products
- Manage conflicts of interest
- Provide clear disclosures and suitable recommendations where required
So AI becomes one more tool inside a regulated framework, not a free pass to automate judgment.
Conclusion
Across buy-side and sell-side firms, AI now supports investment research by cleaning and linking big data sets, reading large volumes of text, extracting patterns across markets, and powering risk and scenario tools—often through internal platforms like Citadel’s AI Assistant, BlackRock’s Aladdin, and other proprietary systems.
Recent work from CFA Institute, IOSCO, the IMF, and industry surveys shows a consistent theme: AI’s main role today is to augment human analysts, not to replace them, while raising new questions about model risk, data quality, and ethical standards.