Hi there,
I have been through the Autocat videos and some of the forum questions here and wanted to better understand how smart is Autocat out of the box? If I am tracking “Coffee Shops” as a category, do I need to create a rule like: Contains "Coffee", categorize as "Coffee Shops"?
Or is Autocat smart enough to handle these?
The reason I ask is that each week I use Tiller, I am spending quite a bit of time creating new rules based on 90 day suggestions. I am also ending up with duplicates since it does not seem like all the rules are being run, so I am trying to get more and more specific. For example:
|Category|Description Contains|
|Dividends|Dividend|
|Dividends|Dividend reinvest Fdrxx Hsa|
|Dividends|Dividend Received Fidelity Treasury Money Market Fund|
Credit Card Payments Payment Thank You - Web
Credit Card Payments Online Payment, Thank You
Credit Card Payments Internet Payment - Thank You
Credit Card Payments Payment Thank You-mobile
Credit Card Payments Mobile Payment - Thank You
Credit Card Payments Online Payment - Thank You
I would have thought that just having Dividend or Payment Thank you would be enough since it is looking for CONTAINS and not absolute string values but it does not seem to be the case…
The time is longer than I would like partly because I think some of these rules are fairly straightforward and partly because Autocat is taking a long time to run (4-5 minutes) on a MacBook Air M2 on Safari in Google Sheets.
The default settings of making changes to new transactions only is selected, so I am not sure why it is taking so long, either. It is making the usage of Tiller on Google Sheets quite lacking and I wanted to see if I can reduce my pain point by making AutoCat “smarter” in any way.
No, Autocat does not have AI/learning. It only does what you tell it to do.
I’m repeating your examples here in table format:
Category
Description Contains
Dividends
Dividend
Dividends
Dividend reinvest Fdrxx Hsa
Dividends
Dividend Received Fidelity Treasury Money Market Fund
Credit Card Payments
Payment Thank You - Web
Credit Card Payments
Online Payment, Thank You
Credit Card Payments
Internet Payment - Thank You
Credit Card Payments
Payment Thank You-mobile
Credit Card Payments
Mobile Payment - Thank You
Credit Card Payments
Online Payment - Thank You
“Description Contains” of “Dividend” will match a Description equal to “Dividend” or “Dividend reinvest Fdrxx Hsa” or “Dividend Received Fidelity Treasury Money Market Fund”. i.e. only your first Dividend rule is needed.
“Payment Thank You” will match for:
“Payment Thank You - Web”
“Payment Thank You-mobile”
but, will not match for:
“Payment, Thank You” (because of the comma)
“Payment - Thank You” (because of the dash)
And it will match based on the order in the AutoCat file and stop at the first match.
So, put more specific rules higher in the list, more general lower.
Another option is to try the Category Hint feature (search the forum for this and you will find the instructions) if you prefer not to create your own Autocat rules (though you will have a significant tradeoff for accuracy and specificity). I’d try creating a few more rules and prove AutoCat is working with some simple rules, as what you describe shouldn’t be happening, and I can’t think of any reason why those original rules wouldn’t work (notwithstanding the punctuation issues @Mark.S mentioned).
I wrote this (which is a somewhat technical install at the moment) and use it entirely instead of autocat in my own sheet. It also requires paying Open AI a small amount of money.
Hoping Tiller will just integrate something like this into the extension. At least for me it’s correct 99% of the time.
Thanks for starting this conversation! It looks like you’ve got a handful of good suggestions here. Did any of these help?
If so, please mark one as the solution.
