Stop rewriting the same credit-risk SQL

Type your table names. Get the prompt, the checks, what good looks like and how to present it.

Try a free tool See the paid tools

7 tools free now
No sign-up
Table names only, never customer data

What slows risk teams down

Rebuilt every month

DPD, buckets, roll rates. From scratch, again.

SELECT dpd ...

One prompt holds the logic.

AI guesses your schema

Rows multiply. DPD starts on the wrong date.

3 loans → 9 rows

Your tables and definitions go in first.

No way to check it

The code runs. Is the number right?

4.2% ← right?

Checks to run, and what good looks like.

Benchmarks that do not match

Gross NPA is not bureau 90+.

Gross NPA ≠ 90+ DPD

Source, period, page and definition shown.

Data cannot leave

No customer data in an AI tool.

customer data ×

Table names only.

From table names to a checked result

Four steps. Your data stays in your database.

1 Your tables

loan_masterrepay_schedulerepay_txn

2 The prompt

ROLE senior credit risk analyst
SCHEMA
repay_schedule(loan_id, due_dt)
repay_txn(loan_id, pay_dt)
TASK DPD and bucket per loan

3 Your AI tool

SELECT loan_id,
  GREATEST(0, asof - MIN(due_dt)) AS dpd,
  CASE WHEN dpd > 90 THEN '90+'
  WHEN dpd > 60 THEN '61-90' ...
FROM repay_schedule ...

4 Checked result

Row count reconciles
No negative DPD
10 loans traced by hand

The jobs that eat your week

Each one is a tool.

Free

Roll rates

Are loans getting worse, month on month?

Free

DPD and buckets

Who is late, and by how much?

Stale pulls3.1%
Duplicate pulls0.4%
Score = -11.8%
Null income7.2%

Free

Bureau scrub

Is this data fit to model on?

Low score38
Income proof24
Policy19
Other19

Free

Declines

Why are we rejecting so many?

With sources

Benchmarks

Is our 90+ high or low?

Opens first

Delinquency review

Why did it move this month?

Illustrative figures except the benchmark, which uses published TransUnion CIBIL personal-loan values.

Set it up once. Every tool knows your tables.

Upload a CSV of your schema, or type it once. Your AI tool forgets your table layout between chats. Your profile does not.

My profile

Schedule table: repay_schedule
Payment table: repay_txn
Due date: due_dt
DPD rule: Days since oldest unpaid due date
Buckets: Current, 1-30, 31-60, 61-90, 90+

Set up my profile

Definition sheet
DPD and buckets
Roll rate
Bureau scrub
Rejection analysis
Every paid tool, when it opens

Fewer prompts, more results

No more re-explaining your tables. Every prompt starts with them.

Less time every month

Change a date, copy, paste. The set-up is already done.

Same definitions, every time

DPD, buckets and NPA rule written once, used by every tool.

No data leaves your database.

NamesYou enter table and column names only
You run itThe code runs on your own data
Your toolUse the AI tool your institution approves

Seven tools are free

Try it: change a name, the prompt changesSample names for a DPD query.

Instalment schedule table

Due date column

Payment date column

Your database

146 more tools, in testing

Each opens only after it passes with risk teams.

Join the waitlist Browse the paid tools

Learn the basics

Plain-language guides for credit-risk analysts.

Hear when the paid tools open

Join the waitlist with your role and the tools you want. We write when they open and pick the first five teams for the delinquency pilot.

Join the waitlist

Free tools

Pick a tool, replace the sample names with yours, copy the prompt. What you type stays in this browser.

Paid tools

146 tools, grouped by the job you are doing. None is open yet: each is tested with risk teams first, then opened. Browse what is coming and join the waitlist for the ones you want.

First to open, a pilot for 5 teams

Why did delinquency move this month?

Three tools that work together. You send only totals you calculate yourself, never borrower records.

Join the waitlist and ask for the pilot

Search the paid tools

Benchmarks

A number is only useful if you know what it measures. This is how a benchmark looks inside a tool, with one example you can compare your own figure against.

What every benchmark shows

The source report, its period and the page
The exact definition, with numerator and denominator
The lending segment and lender type
Year-on-year or quarter-on-quarter change where the publisher gives it
Whether it can be compared with yours: comparable, directionally comparable, or not comparable
A box to enter your own number

We never estimate a value. If no source exists, the tool says "No verified benchmark yet".

What is covered today

14 of the 153 tools carry benchmarks, drawn from RBI, TransUnion CIBIL, Equifax, CRIF High Mark and MFIN: bureau delinquency by product, early-bucket portfolio at risk by product and lender type, microfinance delinquency, and bank gross NPA. Publishers release figures months after the period they describe, so a benchmark describes the last published period.

My profile

Upload a CSV of your schema, or type your tables once. Every tool uses them, so you never explain your data to the AI again.

Table and column names only, never customer data. Your profile is saved in this browser. If you sign in with Google, it is also saved to your account so you can use it on another device. Leave a field empty to use the sample name.

About Bucketwise

Bucketwise gives credit-risk analysts at banks and NBFCs a faster start on routine analysis: delinquency, roll rates, bureau checks and approvals. It is built by a credit-risk practitioner.

What it does

You enter table and column names. The tool writes a prompt for the AI tool your team already uses. You run the code on your own data, check it against our validation list, and read it against healthy and warning ranges.

Your data stays yours

You enter table and column names only, never customer data. This page runs in your browser and sends nothing anywhere. What you type is saved in your browser, on your device, and nowhere else.

Use an AI tool your institution approves

Schema names can still be sensitive. Check your information-security policy before you paste a prompt into any AI tool.

Starting points, not verdicts

The healthy and warning ranges are rules of thumb. Calibrate them to your own portfolio. The tools do not give regulatory advice, and you always paste the regulatory text yourself.

Benchmarks with sources

Every figure names its source, period and page. Definitions that cannot be compared with yours are marked. We never estimate a value.

Tested before it opens

The seven free tools are open now. The 146 paid tools are being tested on real table layouts with risk teams, and each opens only after it passes.

Join the waitlist

Tell us your role and which tools you want. We write when they open, and pick the first five teams for the delinquency pilot.

Open the waitlist form

Terms and conditions

Effective November 2026

1. About these terms

These terms apply when you use Bucketwise (the site). By using the site you agree to them. If you do not agree, do not use the site.

2. What the site provides

The site provides prompts, validation checklists, reading guides and benchmark examples for credit-risk analysis. They are general starting points. They are not legal, regulatory, financial, audit or investment advice, and they do not replace your institution's credit policy, model validation or compliance process.

3. Your responsibility for results

You are responsible for how you use any prompt, any code an AI tool writes from it, and any result. Review and test everything on your own data before relying on it. You are responsible for checking your institution's information-security policy before using any AI tool, and for pasting current regulatory text yourself.

4. What you may enter

Enter table and column names only. Do not enter customer names, account numbers, identifiers, personal data or any confidential data of your institution. We do not ask for them and the site is not designed to receive them.

5. Free tools

The free tools are provided as they are, without any promise that they are accurate, complete, suitable for a particular purpose, or available at all times. We may change or remove a free tool at any time.

6. Paid tools and the waitlist

The paid tools are not open. Listing a tool does not promise that it will be released, or fix a date, scope or price. Joining the waitlist is not a purchase and does not create a right to any tool or to a place in a pilot.

7. Benchmarks

Benchmark figures come from the publishers named beside each figure, and belong to them. We show the source, period and page, and mark definitions that cannot be compared with yours. Figures may be revised or may lag the period they describe. Check the original source before you rely on a figure.

8. Our content and how you may use it

The prompts, checklists, guides and design of the site belong to Bucketwise. You may copy and use a prompt for your own work or your institution's internal work. You may not republish, resell, scrape or redistribute the content, or use it to build a competing product, without our written permission.

9. Limits on our liability

To the extent the law allows, Bucketwise is not liable for any loss or damage, direct or indirect, that arises from your use of, or reliance on, the site, its prompts, its code or its benchmarks. Nothing in these terms limits a liability that the law does not allow to be limited.

10. Changes

We may update these terms. The effective date at the top shows the latest version. If you keep using the site after a change, you accept the updated terms.

11. Governing law

These terms are governed by the laws of India. The courts in India have jurisdiction over any dispute about them.

12. Contact

For questions about these terms, write to .

Privacy policy

Effective November 2026

1. Summary

The site does not ask for customer data. The free tools do not need an account. The only personal information we collect is what you choose to give us on the waitlist form, and your name, email and saved profile if you choose to sign in with Google.

2. What you type on the site

The names you enter into a free tool stay in your browser, on your device. The site does not send them to us or to anyone else. You can clear them by using the reset button or by clearing your browser's site data.

3. The waitlist form

If you join the waitlist, you give us the details the form asks for, such as your name, work email, role, institution type and the tools you want. The form saves your answers in a database run for us by Supabase, which stores them on our behalf. Only we can read them.

4. How we use waitlist details

We use them to tell you when paid tools open, to choose pilot participants, and to understand which tools people want. We do not sell them, and we do not share them with others except the service providers we use to run the form and send email.

5. Signing in with Google (optional)

If you sign in with Google, we receive your name and email address from Google and use them only to keep your profile for you. Your saved profile holds the table and column names and the definitions you typed, and nothing else. You can sign out at any time, and you can ask us to delete your account and saved profile.

5a. Analytics and tracking

We do not add advertising trackers or analytics to the pages of the site.

6. How long we keep details

We keep waitlist details, and any account you create, until the paid tools have opened and the waitlist is closed, or until you ask us to delete them, whichever comes first.

7. Your choices

You can ask us to see, correct or delete the details you gave us, or ask to be taken off the waitlist, at any time. Write to .

8. Changes

We may update this policy. The effective date at the top shows the latest version.