What User-First Lending Actually Looks Like

Lending, Reimagined, Part 5 of 7

A person asking for a loan is not necessarily asking for more debt. They may be asking for a way out of expensive debt, a few weeks of breathing room, or help with a situation they do not fully understand.

The product they ask for is where the conversation starts. It should not decide where the conversation ends.

Consider an illustrative case: a 27-year-old product designer wants a ₹2 lakh personal loan to clear two credit card balances. She has an offer already. Her question is whether she should accept it.

Her financial picture adds some context. She holds mutual funds worth more than ₹2 lakh, and her salary has arrived late twice in the last quarter. There may be alternatives to the offer she has received. But discovering investments does not automatically make borrowing against them the right answer.

A loan against mutual funds is still a loan. The eligible amount depends on the assets and the lender’s terms, not simply the portfolio’s headline value. If the pledged investments fall in value, she could have to provide additional collateral or repay part of the balance. Unresolved shortfalls can lead to a sale of investments. SBI’s loan-against-mutual-fund terms, for example, explain the requirement to maintain the margin throughout the facility.

The useful conversation is: let’s compare the costs, the repayment plan, and what you would be putting at risk.

That is the distinction we are trying to build around at Finanjo. The earlier articles examined expensive credit, regulation, the borrower’s experience, and the pressure that growth places on a fintech’s original mission. This article asks what we should build differently.

1. Suitability comes before eligibility

Eligibility asks whether an application meets a lender’s criteria. Suitability asks whether the product makes sense for the person taking it. One cannot substitute for the other.

For the designer in our example, the interest rate matters. So does the reliability of her next salary. So does the reason the card balances accumulated. A one-off expense and a recurring monthly shortfall need different responses.

Before suggesting a product, we want to understand what the money is for, what the borrower can realistically repay, and which alternatives deserve consideration. A repayment plan should leave room for ordinary life, including the expenses that do not arrive neatly on schedule.

Regulation is moving in this direction. RBI’s Commercial Banks Responsible Business Conduct amendments, issued on 15 June 2026 and scheduled to take effect on 1 January 2027, include unsuitable or inappropriate sales within the definition of mis-selling even where explicit consent exists. The suitability provision includes an exception for products the bank’s policy determines are suitable for all customers. These are provisions for the institutions covered by that instrument, not a blanket claim about every lender. Read the reproduced notification.

In the UK, the FCA’s Consumer Duty began applying to open products and services on 31 July 2023. It requires firms to act in good faith, avoid foreseeable harm and support customers in pursuing their financial objectives. The FCA explains the standard here.

For our product, the practical question is straightforward: before asking someone to apply, have we helped them understand whether they should? Sometimes the answer will be a different loan, a smaller amount, or a conversation with an existing lender.

2. Explain why this lender and this loan

A comparison should make the decision easier to understand. The borrower should know why an option is being shown, which lenders were considered, how the platform is paid, and what remains subject to approval. Comparing partner lenders should not be described as searching the whole market.

RBI’s digital-lending framework addresses this directly. Its 2025 Directions require multi-lender platforms to show matching offers and unmatched lenders, provide key cost and repayment information, and present offers objectively. If offers are ranked, the ranking metric must be publicly disclosed. See the RBI Directions.

Credit Karma illustrates the appeal of matching and the need to get its presentation right. It describes using credit information to recommend products, with a commission typically earned when a recommended product is approved. That is its published business model.

In January 2023, the US Federal Trade Commission finalised an order requiring Credit Karma to pay $3 million to settle allegations about misleading pre-approved and approval-odds claims. The FTC said people had been encouraged to apply for offers for which many did not qualify. Read the FTC’s account.

A prediction about approval must not be presented as a promise. Equally, we should not frighten people away from comparing. A rejected application is not itself a repayment default. A lender’s credit check can create an enquiry, which is a different event. CIBIL explains enquiries here.

The outcome we want is a borrower who can explain the choice: this is the amount I need, this is what it will cost, these are the alternatives, and these are the conditions I am accepting.

3. Understand the whole financial picture, including what the data misses

A credit card, a personal loan and an investment account may sit with different institutions. They still belong to one person with one monthly budget. A card bill cannot be assessed in isolation from that person’s other commitments.

Account Aggregator can help eligible financial information users obtain information from participating institutions with the customer’s consent. Credit reports and separately authorised statements may add other parts of the picture. AA alone should not be described as supplying every card transaction, liability or household commitment.

The advisor still needs to ask questions. An apparent salary payment could include reimbursement. Money transferred by a relative may not be recurring income. A mutual fund could be earmarked for a parent’s treatment. A household obligation may not be visible in the connected account at all.

For Jo, our AI assistant, the standard should be to show what is known, identify what is missing, and ask when the missing information could change the recommendation. More data should improve the conversation, not give us an excuse to stop listening.

The same principle applies to access. The AA consent process identifies the information requested, its purpose, frequency and duration. Users can manage and revoke active consents. Connecting an account should not be treated as unrestricted permission to reuse financial history. Sahamati’s guide explains these controls.

4. Stay useful after the money arrives

Disbursal is a transaction milestone. It is not proof that the borrower’s problem has been solved.

For a consolidation loan, we would want to check whether the intended card balances were actually cleared. For a temporary cash-flow need, we would want to know whether the repayment plan still works. For a securities-backed loan, the collateral obligations continue throughout the facility.

Ongoing support should mean a useful intervention when something relevant changes, based on information the user has authorised us to access. A confirmed repayment falling before expected income could prompt a reminder to check the payment account. A borrower who reports difficulty could be helped to contact their lender early and understand the options it actually offers.

We should not promise that AI can predict every missed payment. Nor should it automatically recommend another loan whenever a cash-flow gap appears. A useful system must recognise when the next step needs human judgment.

RBI’s June 2026 commercial-bank amendments also introduce a mechanism to seek feedback within 30 days of a sale, from their scheduled January 2027 commencement. They allow random sampling and calls or surveys by a function not associated with sales. That requirement is distinct from our proposed ongoing support. Our aim is to remain useful through repayment, including when there is no new product to sell.

5. Build a business that can afford an honest no

It is easy to promise unbiased guidance. It is harder to preserve it when a team has targets, acquisition costs are rising, and one lender pays more than another.

Ditto offers a relevant comparison in insurance. Its published methodology acknowledges insurer-paid commissions and says its evaluations do not depend on payout tables. That is the company’s stated approach, rather than an independent audit of every recommendation. Its evaluation framework is public.

The question is whether the business has controls that keep its commercial interests from deciding the recommendation. For Loan Buddy, the model we want is salaried advisors whose work is reviewed for suitability, explanation and customer outcomes, rather than rewarded solely for disbursals. Recommendations should be documented. Cases that end without a loan should be reviewed for quality too.

The company-level conflict still exists when the individual advisor has no commission target. We need to acknowledge it, disclose relevant relationships, and keep commercial payouts out of the logic used to recommend and rank products.

Our regulatory role must also be clear. Investment-adviser registration relates to investment-advisory activities; it should not be presented as blanket permission for lending or insurance. The applicable framework has to be understood for each service.

In digital lending, RBI’s Directions require fees payable to a lending service provider to be paid by the regulated lender, not separately collected from the borrower by that service provider. Calling a service user-first does not remove that obligation.

What this costs, and what we need to prove

Some consultations will not produce a loan. A smaller loan may generate less commission. A detailed review takes time, and a reliable AI-assisted service still needs engineering, supervision and customer support.

We should account for those costs honestly. Better matching might also help a suitable borrower who would otherwise have applied in the wrong place. The goal is fewer unsuitable loans, not fewer loans regardless of need.

Our commercial hypothesis is that useful guidance will earn repeat relationships and referrals while reducing avoidable mistakes. That is something we have to measure.

Alongside revenue and disbursals, we want to measure whether recommended debt replacements were completed, whether costs were lower after fees, whether repayment remained manageable, and whether users understood the trade-offs.

A smaller monthly payment is not enough evidence of a better outcome. Extending a loan can lower each payment while increasing total interest. A comparison needs to show both.

For the person in our opening example, success would mean helping her make a decision she understands and can reasonably carry through. That is what we want Loan Buddy and Jo to be judged on as we build Finanjo.

The next article looks at what regulation already requires, what the announced changes will introduce, and what policy could do next.

Would we make the same recommendation if this transaction paid us nothing?

The opening scenario is illustrative. Finanjo’s proposed operating practices are commitments to build and measure, not claims of proven outcomes. Regulatory references were checked on 6 October 2026; the June 2026 notification was reviewed through a full-text reproduction.

Previous: The Fintech Failure: Why Digital Didn’t Mean Better
Next: The Regulator’s Dilemma: What Policy Could Do Next
Explore the Founders Office