A borrower asks for their balance at night, and nobody answers until Monday.
An agent answers balances, due dates, payment references and statements at any hour, with the number from your system. Complaints go to your specialized customer unit (UNE) with a case number and the full history.
This is how we would do it: today we don’t run borrower support for any client. What does run is the same kind of agent: Luz handles our own sales in production, and an assistant tested with real messages hands off to a person on a legal threat or a discount request.
What it looks like from customer service.
-
The same questions, all day long.
Balance, due date, payment reference, statement. Your team answers them one by one.
-
After hours, nobody answers.
A borrower who wants to pay on Friday night waits until Monday, and sometimes doesn’t pay.
-
A complaint takes a week to assemble.
The history is scattered across messages, the system and the collection agency.
-
The UNE deadline runs with nobody watching.
A complaint that stays in a chat has no case number and no response date.
What would the agent do?
It would answer frequent questions with the number from your loan system: balance, next due date, payment reference, statement. At any hour, and through the channels your borrowers already use. When someone wants to file a complaint, the agent opens a case number, gathers the full history and hands it to your UNE with the deadline in plain view.
What wouldn’t it do?
It doesn’t resolve complaints or negotiate: that belongs to your UNE and your team. It doesn’t share a loan’s information with anyone other than the borrower. On a discount request, a legal threat or a vulnerable person, it hands the conversation to a person with the full history.
- The agent
- A synthetic agent connected to your loan system read-only and to your messaging channels. It verifies the borrower before sharing any data, answers, opens cases and routes them. It always introduces itself as your lender’s agent.
- The accountable person
- The head of your UNE owns the agent, by name. For the first 4 weeks, the agent checks with a person on every decision and action. After that, anything high-stakes —every complaint, every discount request, every sensitive case— always goes through a person; simple things, like giving a due date, it handles with what it learned, and every answer is logged.
- The evidence
- Every conversation is stored verbatim, with date, channel and what was answered. Every case carries its history and its deadline. It’s auditable: a complaint gets answered with the actual conversation, not with what someone remembers.
- The stop button
- Your team pauses the agent, or a single channel, with one button, and the conversation moves to a person. If the loan system doesn’t respond, the agent doesn’t make up a balance: it says it can’t provide it and routes the request.
Every complaint has a response deadline.
The specialized customer unit answers within a deadline set by law.
Mexican lenders must have a specialized customer unit (UNE), and the law sets a deadline of up to 30 business days to answer inquiries and complaints. A complaint that comes in through a chat at eleven at night counts too. Giving it a case number and a date on the spot is what makes the deadline possible to meet.
Source: Law for the Protection and Defense of Financial Services Users, Article 50 Bis.
We don’t have a borrower support case yet.
We don’t run borrower support for any client, and we won’t say otherwise. What you can check is the kind of agent: Luz, InnovaBlack’s Head of Sales, handles our own sales in production, with a person accountable for her. Write to her below.
- Tested with real messages: a messaging assistant writes formally, doesn’t make up numbers and hands off to a person on a legal threat or a discount request.
- In production at two lenders: the portfolio every answer would come from ties out to the system of record every night.
Talk to Luz. Right now.
Luz runs InnovaBlack's sales conversations — not a recorded demo, the same one that answers our clients. Write to her the way you'd write to anyone.
Prefer WhatsApp? +52 33 3246 1111
Luz
Head of Sales · InnovaBlack
Tell me what your operation needs and I’ll point you to where to start.
If we’re not the right fit, she’ll tell you.
What a lender asks before starting.
- Can an AI agent handle borrower support at a lender?
- Yes, for the repetitive part: balances, due dates, payment references and statements, with the number from your system and after verifying the borrower. Anything that needs judgment —a complaint, a discount request, a sensitive case— goes to a person with the full history. The agent always says it’s an agent and never makes up data it doesn’t have.
- How does the agent work alongside the UNE?
- The UNE still resolves. The agent opens a case the moment a borrower complains, gathers the full history and hands it to the unit with its deadline in plain view. No complaint stays in a chat without a response date, and the UNE gets the case assembled instead of hunting for it.
- What happens if someone other than the borrower asks about a loan?
- They don’t get any information. The agent verifies the borrower before sharing any data, using the criteria your team sets. If it can’t confirm identity, it doesn’t give the balance or the reference and offers the lender’s service channels instead. Every attempt is logged.
- What does it cost, and how do we start?
- You start with a free diagnostic: we look at what your borrowers ask, through which channels, at what hours and how a complaint reaches your UNE today, and tell you which parts an agent can answer. Pricing comes in a formal proposal. There is a one-time activation fee and a monthly operations fee.
Tell Luz what your borrowers ask.
Luz, InnovaBlack’s Head of Sales, asks what she needs and sets up the free diagnostic. If your support doesn’t need an agent, she’ll tell you.