Digital lending in India has changed fast. In just 18 months, a lot has moved.
The Unified Lending Interface (ULI) is now live. More loans now use the Account Aggregator NBFC rail. New co-lending rules began on 1 January 2026. And RBI AI lending rules are taking shape.
These are not just tech topics. They change how an NBFC finds customers. They change credit risk. They change how it works with banks. And they put the board on the hook for loans that a machine approves.
Here are five questions every NBFC board should ask in 2026. Each one comes with the data you need.
The Big Picture: Key Numbers for 2026
| What we track | Latest number |
| Lenders on the Unified Lending Interface (ULI) | 36 in Dec 2024 → 64 in Dec 2025 → 89 in Feb 2026 |
| Banks and NBFCs on ULI (Dec 2025) | 41 banks and 23 NBFCs |
| Data services on ULI | 136+ services |
| Loan journeys on ULI | 12 |
| AA loans in FY25 | ₹1.07 lakh crore, 1.22 crore loans |
| AA loans in H1 FY26 | ₹1.47 lakh crore, 1.5 crore loans |
| AA loans in full FY26 | ₹3.82 lakh crore, 3.68 crore loans |
| AA share of retail and MSME loans (FY26) | 8.4% by value, 11.8% by count |
| AA loans as share of GDP (FY26) | Over 1% |
| AA consents done | 500 million+ (Sept 2026) |
| Bank accounts that can use AA | 2.61 billion+ (early 2026) |
| Accounts linked to AA | 180 million+ (early 2026) |
| First-time borrowers in AA loans | 18.2% by count |
| Women in AA loans | 19.8% by count, 19.1% by value |
| Banks’ share of AA loan value (H2 FY26) | 47.3% |
| Home loans and LAP via AA | Up 624% in one year |
| Average AA loan size (FY25) | About ₹88,457 |
| Co-lending rules start | 1 January 2026 |
| Minimum share each co-lender keeps | 10% (was 20%) |
| RBI FREE-AI report | 7 Sutras, 6 pillars, 26 ideas (13 Aug 2025) |
Sources: RBI and RBIH data (via Business Standard); Sahamati AA Impact Reports, H1 and H2 FY26; RBI Co-Lending Directions, 2025; RBI FREE-AI report.
The message is simple. The data rails are built. Use is growing fast. So the real question is not “should we join?” It is “how fast, how safe, and how do we make money?”
Question 1: Are We on the Unified Lending Interface (ULI)?
What ULI is
The Unified Lending Interface (ULI) is a public data rail for loans. It started as a pilot in August 2023. It got its new name on 26 August 2024. The RBI’s tech arm, RBIH, runs it.
Think of it like UPI, but for loan data. It does not move money. It moves data, with the customer’s OK.
With one link, a lender can pull land records, GST data, PAN and ID checks, dairy and farm data, and AA data. Without it, a lender must build many separate links. That costs time and money.
How fast it is growing
The numbers are strong:
- Dec 2024: 36 lenders
- Dec 2025: 64 lenders (41 banks, 23 NBFCs), up 78%
- Feb 2026: 89 lenders, as per RBIH
- Data services: 136+, up from about 50
- Loan journeys: 12, such as farm (KCC), dairy, MSME, home, personal and two-wheeler loans
What is holding it back
The Unified Lending Interface (ULI) is not a magic switch. Bankers say some land records are still not digital. Some big lenders are also slow to join. So the gains are biggest in states and loan types where the data is clean.
What the board should ask
- Are we live on the Unified Lending Interface (ULI)? If not, when, and at what cost?
- Which of the 12 loan journeys fit our products?
- How much faster will loans get approved?
- How much will we save on paper and checks?
- Who signs off on using ULI data in our credit models?
For a rural or MSME lender, the Unified Lending Interface (ULI) can cut checks from days to minutes. For a city-based lender of small personal loans, the gain may be smaller. Ask for a plan for each loan type. Do not accept a vague “go digital” plan.
Question 2: How Much of Our Book Uses Cash-Flow Lending?
What cash-flow lending means
Cash-flow lending means you lend based on real money flows. You look at what comes into a bank account. You look at GST sales. You look at bills paid on time. You do not rely only on a credit score or on property.
Why the AA rail matters
The Account Aggregator NBFC rail is now the main engine of cash-flow lending in India. An Account Aggregator NBFC is a special NBFC licensed by RBI. Its only job is to move financial data with the customer’s consent. It does not lend. It does not store the data. Well-known names include Finvu, OneMoney, CAMS Finserv and Anumati.
What the data shows
Look at how fast AA lending has grown:
| Period | Amount | Number of loans |
| FY25 | ₹1.07 lakh crore | 1.22 crore |
| H1 FY26 | ₹1.47 lakh crore | 1.5 crore |
| Full FY26 | ₹3.82 lakh crore | 3.68 crore |
That is a jump of more than 3.5 times in value in just one year. AA loans now make up 8.4% of retail and MSME loans by value. By count, they make up 11.8%. That means AA loans tend to be smaller and more frequent.
NBFCs led this change. But banks are catching up fast. In H2 FY26, banks made 47.3% of AA loan value.
Three more facts stand out:
- It helps new borrowers. First-time borrowers made up 18.2% of AA loans by count. Women made up 19.8% by count.
- It is moving into secured loans. Home loans and loans against property via AA grew 624% in a year. So cash-flow lending is no longer just for small personal loans.
- The rules are getting stronger. In June 2026, RBI named Sahamati as the self-regulatory body for AA. By September 2026, AA had crossed 500 million consents.
What the board should ask
- What share of our loans use Account Aggregator NBFC data?
- How does this differ by product?
- How many customers finish the AA consent steps? Where do they drop off?
- Which Account Aggregator NBFC partners give us the best success rate?
- Do our cash-flow lending models beat score-only models on early defaults?
- Do we use GST and bank data for MSME cash-flow lending, or only for personal loans?
If these numbers are not in your board pack, you are flying blind. Cash-flow lending is now a basic need. The edge comes from how well you use the data.
Question 3: Is Our AI-Led Lending Ready for RBI Checks?
The RBI’s AI plan
RBI AI lending rules are being shaped by a report called FREE-AI. The full name is the Framework for Responsible and Ethical Enablement of AI. RBI set up the panel in December 2024. The report came out on 13 August 2025.
It has three parts:
- 7 Sutras (core values): trust first, people first, new ideas over fear, fairness, clear ownership, easy-to-explain systems, and safety.
- 6 pillars: tech base, policy, skills, governance, customer protection and checks.
- 26 ideas to put the plan into action.
The panel also suggested a test space (a “sandbox”) for AI. It suggested shared computing power, so small lenders are not left out. It also proposed a public fund to back home-grown AI models.
What this means for NBFCs
RBI AI lending rules point to six clear steps:
- A board-approved AI policy. Write down where you use AI, how much risk you accept, and who is in charge.
- Clear reasons for every “no”. The report links the Digital Lending Directions, 2025, to telling borrowers when AI judged their loan. A rejected customer should get a reason they can understand.
- A list of all models. Track every AI model in use. Note its data, test results and any drift.
- Bias checks. Check that approval rates and prices are fair across gender, region and other groups.
- Clear blame when things go wrong. Know who owns each AI decision. Report AI mistakes.
- Use the sandbox. Test new AI tools in a safe space first.
Many RBI AI lending ideas are still advice, not hard rules. But boards should not wait. Checks will focus on how the board oversees machine-made loan decisions. This matters most when AI models use Unified Lending Interface (ULI) and Account Aggregator NBFC data at scale.
What the board should ask
- Do we have an AI policy that fits RBI AI lending values?
- Which loan decisions are fully automatic?
- Where does a human still check the decision?
- Can we explain each “no” to a customer and to RBI?
- When did someone outside the team last test our models for bias?
Question 4: Does Our Co-Lending Model Follow the 2026 Rules?
The new rules
RBI issued new co-lending rules on 6 August 2025. They began on 1 January 2026. They replace the old 2020 rules, which only covered priority sector loans. Now the co-lending model can cover all types of loans.
Old rules vs new rules
| Topic | 2020 rules | 2026 rules |
| Which loans | Priority sector only | All loans |
| Share kept by each lender | 20% (NBFC) | 10% each |
| Picking only some loans (“cherry-picking”) | Allowed | Not allowed under co-lending |
| Time for partner to book its share | Loose | About 15 days |
| Data sharing | From time to time | By end of next working day |
| Bad loan (NPA) tagging | Each lender’s own way | Same for the borrower across both lenders |
| Public disclosure | Little | Partners and loan data on website and in accounts |
| Interest rate | Varied | One blended rate for the borrower |
Why it matters
The lower 10% share is good news. It lets an NBFC lend more with the same capital. That makes the co-lending model great for growth with less capital.
But the new rules are strict on process. Data must flow by the next working day. Bad loans must be tagged the same way by both partners. This needs strong tech links with bank partners.
This is where it all comes together. A co-lending model built on Unified Lending Interface (ULI) data, Account Aggregator NBFC consent and clear AI scores is easy to defend. A co-lending model built on PDFs and manual checks is not. Digital co-lending must also follow the Digital Lending Directions, 2025. So both sets of rules apply.
What the board should ask
- Do all new co-lending model deals since 1 January 2026 follow the new rules?
- What return do we earn on the co-lending model versus loans on our own books?
- Can our systems share data by the next working day?
- Can we tag bad loans the same way as our bank partner?
- Do our bank partners accept our cash-flow lending approach?
Question 5: What Is Our 2-Year Digital Lending Plan?
The last question ties it all together. A good plan links three things:
- Data rails: ULI and AA
- Decisions: AI-led credit checks
- Reach: co-lending and partners
A simple board scorecard
| What to track | Why it matters |
| % of loans using Unified Lending Interface (ULI) or AA data | Shows how much we use the new rails |
| Average time to approve a loan | Shows the speed gain |
| AA consent success rate | Shows where customers drop off |
| Early defaults: cash-flow lending vs score-only | Proves the value of new data |
| Share of first-time and MSME borrowers | Tracks reach into new markets |
| Co-lending model loan book and return | Shows if low-capital growth works |
| AI models tested in the last 12 months | Proof of RBI AI lending oversight |
| Complaints about machine decisions | Early warning on fairness |
A simple timeline
- 0 to 6 months: Join ULI. Use AA for all products. Pass an AI policy.
- 6 to 12 months: Put cash-flow lending models live. Link systems with co-lending model partners. Set up model risk checks.
- 12 to 24 months: Grow secured cash-flow lending, like home loans and LAP. Add more ULI journeys. Get ready for firm RBI AI lending rules.
Key Takeaways for NBFC Boards
- The Unified Lending Interface (ULI) is live and growing fast. Early movers will win in rural, farm and MSME loans.
- Each Account Aggregator NBFC link gives you richer data. The edge comes from how you use it.
- Cash-flow lending has grown from small personal loans to MSME and home loans.
- The new co-lending model rules reward NBFCs with good data and strong systems.
- Under RBI AI lending rules, the board, not just the team, is responsible for machine-made loan decisions.
Let’s Connect
Want help to build your ULI, AA, RBI AI lending or co-lending plan for 2026? Connect with the NBFC Advisory team today, and let’s make your NBFC ready for the next wave of digital lending.