https://discord.gg/botpress logo
Join Discord
Powered by
Product Recommendation System Issue
# 🤝help
s
So, i've created a product recommendation system bases on a table which contains all of the products and their different chacacteristics. It follows this system: A Raw input card asks the user what where they looking for in a certain category that was proviously chosen by them. (Each category has its own table with their respective products). The user answers and is transitioned to another node, where a find records card is triggered to look in that same table. It has no filters and the results are saved in a variable that just for the sake of this post we're gonna call it "x". "x" contains all of the products and their characteristics in that specific category table, since no filter was applied. After that, an ai task which is prompted to analyze the user's request and create 3 recommendations based on that "x" variable (since it contains all of the table's info) is triggered and it works perfectly fine for me because the recommendations are accurate. The recommended products are then saved into a "y" variable. Till now everything works perfectly fine but when I try to deploy these reommendations into the chat with a carousel card (which is supposed, in this case, to contain the name of the product, the price, and an image of it) the problem happens. None of the carousel cards contain neither the name, the price nor the image of each product and I really don't know why this is happening. I will send images so you guys can understand better.
here is a picture of what the cards look like
Here is the "y" variable which is the "workflow.productorecomendado"
then the [1] contains the product that was recommended, like in this case it should show the information of the second product that the ai task came up with
and the final part that says .Precio or .Nombre are supposed to be the name and the price of the product, both having a specific column for each of the products in the table
here is a screenshot from the event debugger
f
Hey there Matt Take a look at this video and let me know if it helps.

https://www.youtube.com/watch?v=V1D34HPhCd8▾

If it doesn't help: Looks like the problem is with workflow.productorecomendado. Can you take a screenshot of the variable and what information it contains?
s
It worked
but now I have another problem lol
It'll always recommend me the same products with the same prompt
for example
if there are 4 black t-shirts and I ask it to recommend me black t-shirts it'll always send me to the same 3
b
for the recommending, we cache ai requests to minimize the amount you spend on ai tokens
so if you're prompting the bot with the same thing it'll return the same results
because in theory your users won't be asking the same questions over and over
s
Thanks Robert!
And I have another question
If the client asks for a product that doesn't exist
for example a red t-shirt when we don't have one
how do I make a "fallback" for that
without sending the carrousel
Just Solved It
lmao
b
hahahaha
too fast for me
how'd you solve it?
just curious!
s
Used an expression card so that when the length of the variable where the recommended product was stored was detected as 0, it would send the fallback message
but again I'm having another problem
made another table but it has more than 20 items
but the find records card can't detect all 20
idk why
b
hmm
it can't detect all 20 or ones beyond the first 20?
s
ones beyond the first 20
sorry expressed myself bad lol
any idea why, robert?
@bumpy-butcher-41910
sorry for tagging but I really need a fix asap
b
sorry matt, I can't explain why your table wouldn't be pulling results beyond the first 20
s
not the table
My table has 26 items. I used a find records card with no filter for that same table so it should save to the created variable thos 26 items
but it only saves 20
b
can you load a report id and send it here?
s
done
00091c7c-9056-4d52-a350-f369f7e748f9
here my report id
PreviousNext
Powered by