In partnership with
Hey, it’s Jeffrey — back again!
I keep a pretty close eye on my subscriptions. I know roughly what I'm paying for and I cancel things I don't use.
What I had never done was check whether the subscriptions I kept got more expensive over time. Streaming services raise prices quietly. Software subscriptions tick up a dollar or two a year. You approve it once, by inaction, and then never think about it again.
So I pulled a transaction export and asked AI to find every place a recurring charge had crept up since I first signed up.
Here's the prompt I used:
I want to check whether any of my recurring subscriptions have gotten more expensive over time without me noticing.
Below is a transaction export from my bank or finance app (Monarch, YNAB, or similar) covering recurring charges. Please review it and do the following:
1. Identify recurring subscription-like charges (same or similar merchant name appearing monthly or annually).
2. For each one, check whether the charge amount has changed over the time period shown. If it has increased, calculate the dollar and percentage increase from the earliest to the most recent charge.
3. Total my current monthly subscription spend, and separately total the combined increase across any subscriptions that have crept up in price.
4. Give me a ranked list: which subscriptions have increased the most in dollar terms, and flag anything where the increase looks unusually large relative to the original price.
5. Note any recurring charges where the merchant name is unclear, so I can identify what they are.
Keep the output simple and scannable. Only report changes you can actually see in the data, do not guess at anything not shown.
[PASTE TRANSACTION EXPORT BELOW OR ATTACH FILE]
Here's what came up when I ran it on mine.
Based on my Monarch data, I had 24 active subscriptions totaling about $460 a month. Most of them hadn't moved in price at all. But three had crept up since I first signed up, adding $61.83 a month combined, and one annual subscription was running about $21 a year higher than when I started it.
The two largest jumps had both more than doubled from their original price. Neither one sent an email I remembered seeing. I just kept paying the new amount without noticing the gap between what I signed up for and what I'm actually charged now.
There was also a smaller, unexpected find. One of my recurring charges showed up on my statement under a merchant name that didn't match the service at all. AI flagged it as worth checking, and it turned out to be a subscription I knew about, just labeled in a way my bank made unrecognizable.
A few subscriptions I thought were still active hadn't charged me in months, which is its own kind of useful to know even though it wasn't what I was looking for.
One practical note: redact account numbers or any other sensitive info before pasting a transaction export anywhere. Merchant names and amounts are all AI needs to do this analysis.
Keep going? Want to go further? Try these follow-up prompts after you get your initial result:
If you want to dig into the unclear charges: "Here are the recurring charges you flagged as unclear: [list]. Based on typical naming conventions banks use for these merchants, what services might these actually be?"
If you want to check for cancelled-but-still-active subscriptions: "Looking at this same export, are there any subscriptions that charged regularly for a while and then stopped, which might mean I cancelled them, or might mean they are about to lapse for a different reason? Flag anything with a gap of 3 or more months."
If you want a renewal calendar: "Based on the charge dates in this export, build me a simple list of which subscriptions renew in which month, so I know when to expect each charge."
AI tip worth trying this week: cross-check high-stakes AI answers
I got a wedding tip recommendation from an AI too, and I almost just went with it. Instead I ran the same scenario by a second AI as a gut check, and it flagged that the first answer had applied the wrong context entirely. That mismatch was the signal to actually go research it myself instead of trusting either answer blindly.
The move: run high-risk AI answers through a second AI before you act. You're not looking for a final answer, you're looking for a disagreement. If the two line up, that's a decent sign. If they don't, that's your cue to dig in further.
f you work in fintech or finance, you already have too many tabs open and not enough time.
Fintech Takes is the free newsletter senior leaders actually read. Each week, I break down the trends, deals, and regulatory moves shaping the industry — and explain why they matter — in plain English.
No filler, no PR spin, and no “insights” you already saw on LinkedIn eight times this week. Just clear analysis and the occasional bad joke to make it go down easier.
Get context you can actually use. Subscribe free and see what’s coming before everyone else.
One quick note: This newsletter is for educational purposes only and does not constitute financial advice. I'm not a financial advisor — just someone sharing ideas and tools I've found useful. Use what works for you, skip what doesn't, and always do your own research. Some links may be affiliate links or sponsored content for which I may receive compensation.