Obviously AI perfectly handles all semi-structured and un-clean data.
Don't worry about spending hours (or waiting on your engineers) to get clean and pre-processed your data. Obviously AI works perfectly fine with semi-structured and un-clean data. Everything from data cleaning to insight is automated. That being said, your data needs to have 3 basic requirements as noted below.
Your dataset size should be at least 10,000 rows x 10 columns.
If you are connecting a database, please make sure it has an identifier column. This means a column that has some form of primary key, example a column with user names, ids, emails, or just plain numbers that can identify one row from another.
Please make sure your column names are humanly read-able. Meaning, column names like "Customer ID" (or even misspelled versions like "cstmr_id") are permitted. However, column names like "ABC123" may not yield the best results.