How should an investor relations team learn to use AI?
It depends on how much of your IR work is writing versus data. If most of the week goes to LP updates, DDQ answers and meeting prep, a short hands-on course built on your own past documents fits best, followed by a shared prompt library. If the pain is pulling numbers from fund admin reports and portfolio data, training alone will not fix it, and you will also need a connected workflow or agent. Either way, train on real IR tasks in your firm’s approved tools, never generic demos.
Who this is for
Heads of investor relations, IR associates, fund CFOs and COOs at private equity, venture, credit and other private capital firms, plus anyone at a company who owns shareholder or investor communication.
The short version
- Start with the three or four tasks that eat the most IR hours, usually DDQs, quarterly letters, meeting prep and inbound questions.
- Teach people to work from approved source documents, not from memory or the open web.
- Set clear rules on LP and fund data before anyone uploads anything.
- Measure hours saved and turnaround time on real requests, not course completion.
Why IR is a good place to start
Investor relations work is mostly reading, summarizing and writing to a high standard, with a lot of repetition. The same questions come back in every due diligence questionnaire. Quarterly letters follow a familiar shape. Meeting prep means pulling the right facts about an LP, their history with the firm and their recent questions. Modern AI assistants are good at these jobs when they are given the right source material and a person checks the output.
IR is also a high-trust function. A wrong number in an LP letter or a DDQ answer that contradicts last year’s is a real problem. That is why training matters more than licenses here: the skill is not “use ChatGPT”, it is “use AI quickly without letting an error reach an investor”.
What to train on
Build the program around tasks your team already does. A practical starting list:
- DDQ and RFP answers. Draft first answers from your approved answer library and past questionnaires, then have a person check every fact against the source. Teach people to ask the assistant to cite which document each answer came from.
- Quarterly letters and updates. Turn portfolio notes and approved figures into a first draft in the firm’s voice. Train people to give the assistant a past letter as a style example and to paste in only the approved numbers.
- Meeting prep. Summarize an LP’s past correspondence, meeting notes and questions into a one-page brief before a call.
- Inbound question triage. Sort and draft replies to routine LP emails, leaving anything sensitive to a senior person.
- Consistency checks. Ask the assistant to compare a new draft against last quarter’s letter or the latest DDQ and flag statements that changed.
Each of these should be practiced on the firm’s own (approved) material, so people leave with prompts and habits they will use on Monday.
Rules for LP and fund data
Before training, agree what can and cannot go into AI tools. LP names, commitments, side letter terms and fund performance are often covered by confidentiality obligations. A simple set of rules:
- Use only the firm’s approved business AI accounts, never personal ones.
- Check side letters and LP agreements for confidentiality terms that affect AI use.
- Treat AI output as a draft. Every figure in an investor-facing document must be checked against the source.
- Keep a record of which documents an answer library was built from and when it was last reviewed.
Our pages on keeping client data safe with AI and using AI on NDA documents go deeper, and the rules belong in your AI acceptable use policy.
What a short program looks like
For most IR teams, a compact format works better than a long course:
- Session one, about 90 minutes: the basics of the approved tool, data rules, and live practice on a DDQ section using the team’s own past answers.
- Session two, about 90 minutes: letters and meeting prep, with each person drafting a real piece of upcoming work.
- Two to four weeks of follow-up: short check-ins where people bring live work, share prompts that worked and fix the ones that did not.
- A shared prompt library: a single page of tested prompts for each recurring task, owned by one person on the team.
See how long it takes to train a team on AI and live training vs self-paced courses for how to choose a format.
When training is not enough
Training helps people draft and check faster. It does not move data between systems. If your IR team spends hours copying figures out of fund administrator reports, the portfolio monitoring tool and spreadsheets, the bigger gain usually comes from an agent or workflow that assembles that data for review. Our guide on automating LP reporting with AI agents covers that path. A sensible order is to train first, learn which tasks repeat, and then automate the ones that are stable.
How to tell if it worked
Pick a few measures before you start and check them a month or two later:
- Time to complete a standard DDQ section or quarterly letter draft.
- Turnaround time on routine LP questions.
- Number of corrections a reviewer makes on AI-assisted drafts.
- How many people on the team use the prompt library each week.
Our page on measuring AI training results explains how to set a baseline without making it a project of its own.
Common mistakes
- Generic demos. Showing how to write a poem teaches nothing about a DDQ.
- No data rules. People either avoid AI entirely or use it carelessly.
- Trusting numbers. AI can produce a figure that looks right and is not. Every number gets checked.
- One-off training. Without follow-up and a shared library, habits fade within weeks.
How Ruby helps
- AI training: hands-on sessions for IR teams built on your own past DDQs, letters and meeting notes, with follow-up coaching.
- AI strategy and consulting: setting data rules for LP information and deciding which IR tasks to automate.
- AI engineering: building agents that pull approved data together for LP reporting and answer libraries that stay current.
If you want your IR team using AI on real work safely, book a call with Ruby.
