Word normalisation before hitting a database

Hi guys, I have a question i am not able to figure out, I hope to get some help.

I am designing a simple bot that will answer hardware brands questions, e.g., Sony or Microsoft. I have a problem with misspelling & normalisation. If the user asks about “sony” I want the system to transform to “Sony”. I am using synonyms to do this but it doesn’t look like a good strategy because I have to list there all possible misspellings for all brands in the DB. Moreover, we use look-up tables to better organise bot’s knowledge. We have a table Vendor were we put all brands.

That means that we do not have a synonym entry for each brand listed in the Vendor look-up table.

To make the problem harder, it is difficult to prepend a spell checker in the pipeline because brands are proper names!

any help is appreciated.