About CardCopilot
The fine print, turned into one clear answer
Indian credit cards are quietly one of the best reward ecosystems in the world, and one of the most confusing. CardCopilot models every card as data — rates, caps, exclusions, milestones, fees — and computes which card to use, which to keep, and what each one is actually worth to you.
- 1,143
- Cards modelled
- 56
- Issuers
- 4,213
- Reward rules
- 79
- Transfer partners
Across the Indian market
Banks and fintech issuers
Rates, caps, exclusions
Airline and hotel programmes
Why this needed building
Accelerated categories, portal multipliers, spend caps, merchant exclusions, milestone benefits and fee waivers all interact in ways no spreadsheet keeps up with. Most people leave real money on the table for one reason: the best card for the purchase in front of them is not obvious, and working it out at the till is not realistic.
So the arithmetic is the product. A deterministic engine evaluates your spend against every card's full configuration — the same inputs always give the same answer — and PRISM narrates the result. The model chooses which calculation to run. It never produces a figure itself.
From a swipe to an answer
Six stages, in order. Nothing in this path guesses.
A purchase
An amount, a merchant, a channel — typed in, or imported from a statement.
Merchant resolved
The raw descriptor becomes a known merchant, a category and an MCC.
Rules matched
Every rule on every card is tested: base rate, accelerated category, merchant multiplier, exclusion.
Caps applied
The cap ledger says what each rule has already paid out this cycle — so the answer is the next rupee, not the advertised one.
Points priced
Earnings are valued by redemption mode, and the assumption behind the number is carried with it.
PRISM explains
The figures are narrated. The model chooses which tool to call; it never produces a number itself.
Why a “5% card” is usually not a 5% card
A cap turns one advertised rate into a blended one. Spend past it and every further rupee earns the base rate, dragging the average down — silently, because the statement never shows you a percentage.
Show the numbers
| Monthly spend | Advertised | You earn |
|---|---|---|
| ₹0 | 5% | 5.00% |
| ₹10k | 5% | 5.00% |
| ₹20k | 5% | 3.00% |
| ₹30k | 5% | 2.33% |
| ₹40k | 5% | 2.00% |
Illustrative: a 5% dining rate capped at ₹10,000 a month, 1% thereafter. Not a specific card — the shape is what matters, and it is the same for every capped rate.
How we hold it
Independent by design
No issuer pays us to rank their card higher. Rankings come from arithmetic on reward configurations — not from who is paying.
Show the working
Every figure traces to a rule: an earn rate, a cap, an exclusion, a milestone, a fee. If we cannot show where a number came from, we do not show the number.
Net value, not headline value
Points are priced at conservative redemption values and fees are subtracted, so a flashy card with a large fee does not beat a quiet free one on paper.
What we will not do
We never store a full card number. We never predict an approval or a credit score outcome. And PRISM refuses to state a figure it cannot trace to a calculation.
What CardCopilot cannot do
- Predict whether you will be approved for a card. Nobody outside the issuer can, and anyone who claims otherwise is guessing.
- Tell you what your credit score will do. Utilisation and history are inputs to models we cannot see.
- Show your credit utilisation. We hold your limits and your logged spending, not your outstanding balance — and cycle-to-date spend is a different number that would be wrong on any card you have already paid down.
- Promise a redemption value. Transfer ratios and award charts change without notice, so every valuation carries the assumption behind it.
Saying so is the point. A rewards platform that will not tell you where its knowledge stops is a rewards platform you cannot check.
Find your best card in two minutes
Answer a few questions, browse the full explorer, or just ask PRISM. No signup required for any of it.