Product documentation checked November 4, 2026. Examples are illustrative; features and availability can change.
Short answer: An AI tool inventory is one page that lists every AI tool your team touches, what data each one can see, where its output goes and who checks that output before it reaches a borrower. It takes about an hour, and it’s the first thing your company will want if anyone asks how your team uses AI.
If you lead a team of four loan officers and a processor, and your branch manager just asked what AI your people use, start here.
Quick test: how many AI tools does your team use?
Now count again, slower. The note-taking app that transcribes your calls. The “help me write” button inside your email. The summary feature your CRM switched on last quarter. The browser extension that cleans up your texts. The chatbot someone on the team uses to draft a market update.
If you’re like most people, the second count is bigger than the first. The tools that only turn up on the second count are the ones nobody approved, and they’re the ones compliance will ask about. You can’t check a tool you forgot you were using.
The belief that catches teams: our company has an AI policy, so we’re covered. A policy says what’s allowed. It can’t list the tools your team actually opened this week.
This is a practical template, not legal or compliance advice. Your company’s compliance team and its written policies decide which tools are allowed and how. Use this page to bring them a clear picture, not to replace them.
Skip this if your company already inventories AI use for your team, down to browser extensions and phone apps. Ask for that list instead.
What counts as an AI tool on this list?
Anything that generates, summarizes, classifies or transcribes content for you. That includes:
- Standalone chat tools (ChatGPT, Claude, Gemini, Copilot and the like).
- AI features inside software you already pay for: your CRM, email, phone system, video calls or document tools.
- Transcription and meeting-note apps, including the ones on your personal phone that you use for work calls.
- Browser extensions and keyboard apps that rewrite or suggest text.
- Anything a vendor describes as “AI-powered,” even if you’ve never clicked the AI part.
When in doubt, list it. It’s easier to cross a tool off than to explain why it was missing.
What goes on the page?
One row per tool. Eight columns. Here’s the template with one example row filled in (the tool is a generic example, not a recommendation):
| Tool | Who uses it | What it does for us | What data it can see | Where the output goes | Who reviews it | Approved by | Last checked |
|---|---|---|---|---|---|---|---|
| Meeting notes app (example) | LO and processor | Transcribes and summarizes calls | Everything said on the call, including borrower details | Summary pasted into CRM notes | LO, before saving | Company compliance (date) | Month, year |
The column that does the most work is “What data it can see.” Be specific. “Call audio including borrower income and employer” is useful. “Some customer info” isn’t.
The column that catches the most problems is “Approved by.” If the honest answer is “nobody, I just started using it,” write that down. That row is your first conversation with compliance.
If you’d rather paste this into a doc or a spreadsheet, here’s a plain-text version:
AI TOOL INVENTORY Team: __________ Updated: __________
Tool:
Who uses it:
What it does for us:
What data it can see:
Where the output goes:
Who reviews the output (before it reaches a borrower):
Approved by / date:
Last checked:
(Repeat one block per tool.)Why are lenders being asked about AI now?
Because the agencies started asking. These rules apply to seller/servicers, the companies that sell loans to or service loans for Fannie Mae and Freddie Mac. As a loan officer, you’ll feel them through your employer’s policies, not as a rule addressed to you personally.
- Fannie Mae: Lender Letter LL-2026-04, issued April 8, 2026 and effective August 6, 2026, says seller/servicers must be prepared to disclose to Fannie Mae, on request, the AI they use, how they use it and the safeguards around it. AI used by vendors is expected to be managed no less protectively (Fannie Mae, LL-2026-04). Note who receives that disclosure: Fannie Mae, not borrowers.
- Freddie Mac: Bulletin 2025-16, dated December 3, 2025, set an AI and machine learning governance framework that took effect March 3, 2026 (Freddie Mac, Bulletin 2025-16).
Your company can’t describe what AI it uses if it doesn’t know. Some of that knowledge lives with the people who actually use the tools every day, which is you and your team.
Lenders are working on this. A survey of 31 lenders and servicers by the MBA, AARMR and BCG (April to July 2026) found 27 of 31 have written AI policies, but only 18 of 31 monitor AI use on an ongoing basis. Regulatory and compliance concerns came out as the biggest barrier to adoption (MBA, AARMR and BCG, published September 2026). So most lenders have the policy, and far fewer keep watching what’s actually in use. A written policy and a current list of the tools in use are two different things. This page helps with the second.
Who should own the list, and how often should it change?
One person on the team keeps it. Not the whole team, one name. Everyone else has one job: tell that person before they start using a new tool, not after.
Review it monthly. It takes ten minutes once the first version exists. Run the second count again: check for tools nobody uses anymore, features a vendor quietly switched on and any row where “Approved by” is still blank.
Then send it to whoever handles compliance or operations at your company, in whatever format they ask for. Your team’s list is an input to their work, not a replacement for it.
What if a tool on the list hasn’t been approved?
Stop putting borrower information into it until your company says otherwise. That doesn’t mean the tool is bad. It means the decision belongs to someone with the authority to make it. Bring the row to compliance with the “What data it can see” column filled in, because that’s the first thing they’ll ask.
Frequently asked questions
Do Fannie Mae’s AI rules apply to individual loan officers?
They’re written for seller/servicers, the companies doing business with Fannie Mae. Loan officers feel them through their employer’s policies. Ask your compliance team how your company applies them.
Do AI features inside tools we already pay for count?
Yes. A summary button inside your CRM or a transcription feature in your phone system sees data the same way a standalone app does. List them.
Should borrower information ever go into a general AI chat tool?
Only if your company has approved that specific tool for that specific use. Without that approval, keep borrower details out.
Compliance note. Automated outreach to consumers and the handling of borrower data are regulated. Have the wording, consent and opt-out handling of any workflow reviewed before it goes live. This article is general information, not legal advice.
