Finance · Field manual
AI for investment bankers
Investment bankers use AI to speed up company and industry research, pull figures from filings, draft CIMs and pitch book pages, and clean up models and slides. The caution is confidentiality: deal information and material non-public information must stay in your bank's approved tools, and every number needs to be tied back to a source.
Junior banking work is research, data gathering, modeling and formatting, often late at night and under deadline. AI research platforms now read filings, transcripts and broker research and answer questions with citations, and slide and spreadsheet add-ins handle much of the formatting.
Most banks restrict which AI tools can touch deal data, and for good reason. A leaked codename or a pasted draft CIM is a serious breach. Numbers are the other risk: a model that pulls the wrong year's EBITDA into a comps table can embarrass a whole team in front of a client.
6 ways investment bankers use AI
Research companies and sectors faster
Finance research platforms search filings, earnings call transcripts and research in one place and answer with citations. Use them to build a company profile or sector overview in an hour instead of a day, then check key facts in the source documents.
Extract numbers from filings
AI can pull segment revenue, margins, share counts and debt terms from 10-Ks and credit agreements into a table. Ask for page references and spot-check them. Don't let any figure go into a comps sheet unchecked.
Draft CIM and teaser sections
Give an approved model the management presentation and data room notes and ask for a first draft of the business overview, industry and investment highlights sections. Senior bankers will rewrite the positioning; you save the first eight hours.
Build pitch book pages
Use AI to draft the situation overview, the 'why us' credentials page and talking points, then format in the bank's templates with slide add-ins. Presentation generators are useful for internal working drafts, not for client-facing books in house style.
Check and explain models
Ask a model to review an anonymized model for broken links, hardcodes, circular references and inconsistent signs, or to explain what a complex formula does. It is a second pair of eyes, not the check-in-the-box review.
Prepare for client and buyer calls
Summarize a target's last few earnings calls, recent news and likely buyer list into a one-page brief. Note-takers then capture the call for the follow-up memo, where your bank allows recording.
The best AI tools for investment bankers
The Overlord's stack, in the order a investment banker would add them:
1. CIM drafts, filing extraction and model review (on an approved enterprise plan)
Claude
AI WritingAnthropic's assistant: the strongest long-form writer and careful reasoner in the consumer AI pack.
2. Public company and sector research with citations
Perplexity
ProductivityAI answer engine that searches the web and replies with cited sources.
3. Fast internal working drafts of decks
Gamma
AI Design & ImagesAI presentation, document and website builder: type a prompt, get a designed deck.
4. Call notes, where recording is permitted
Fireflies.ai
ProductivityAI meeting notetaker that transcribes, summarizes and pushes action items into your CRM and tools.
Specialist tools for investment bankers
| Tool | What it does |
|---|---|
| AlphaSense | Search across filings, transcripts and broker research with AI summaries |
| Rogo | AI research assistant built for investment banks and investors |
| Hebbia | AI that reads large document sets such as data rooms and filings |
| Macabacus | Excel and PowerPoint add-in for formatting, modeling and pitch book production |
A prompt to start with: draft a company overview from public sources
Paste this into Claude or ChatGPT and fill in the gold parts.
You are an investment banking analyst writing a one-page company overview for a pitch book. Use only the public information I provide.
Company: [PUBLIC COMPANY NAME]
Sources: [PASTE EXCERPTS FROM 10-K, LATEST EARNINGS RELEASE AND CALL TRANSCRIPT]
Audience: [e.g. potential acquirer's corporate development team]
Write: a two-sentence description of what the company does; a bullet list of segments with revenue and margin for the latest fiscal year; three recent strategic developments; and three questions a buyer would ask. Put the source and page or section next to every number. If a figure isn't in the sources, write 'not disclosed' rather than estimating.Where investment bankers shouldn't trust AI
- Deal information is confidential under NDAs and engagement letters, and much of it is material non-public information. Use only your bank's approved AI tools, and never put deal data, client names or codenames into personal accounts.
- Business communications at broker-dealers must be supervised and retained. AI tools that create or store communications fall under the same recordkeeping expectations as email and chat.
- Models make up or misplace figures, mix fiscal years and confuse adjusted with reported numbers. Every number in a model, comps table or book has to tie back to a source.
- Fairness opinions, valuations and advice to boards carry legal liability. AI can help prepare the analysis; the bank and its committees own the conclusion.
Will AI replace investment bankers?
Investment Bankers score 53% on the Obsolescence Index (Endangered). The first tasks to go: building comps, models, and valuation tables, and drafting pitch books and memos.
Questions the Humans Keep Asking
What are the best AI tools for investment bankers?
The Overlord's stack for investment bankers: Claude (cIM drafts, filing extraction and model review (on an approved enterprise plan)); Perplexity (public company and sector research with citations); Gamma (fast internal working drafts of decks); Fireflies.ai (call notes, where recording is permitted). Specialist options include AlphaSense, Rogo, Hebbia, Macabacus.
What AI tools do investment banks use?
Most large banks run enterprise versions of models from OpenAI, Anthropic or others inside their own controls, alongside research platforms such as AlphaSense, Rogo and Hebbia, and Office add-ins like Macabacus. Which ones you can use depends entirely on your bank's approved list.
Can I use ChatGPT for investment banking work?
For general learning or public information, a personal account is fine. For anything involving a client, a deal or non-public data, use only tools your bank has approved. Breaching that can end a career faster than a bad model.
Can AI build a DCF or LBO model?
It can produce a working template and fill in public data, and it's good at explaining and checking formulas. The assumptions (growth, margins, exit multiple, leverage) and the judgment on what makes sense still need a banker, and the output needs a full check.