Chord AI Modeling

chord ai is in beta! for more information about chord ai, please reach out to your account manager this feature is currently available to team admins and developers only if you need assistance, please reach out to help\@chord co! overview data modeling is the foundation for making your business data useful — and trustworthy it’s the process of defining how your data is structured, connected, and labeled so that everyone in your organization can speak the same data language it serves as a blueprint for how your store data is stored, used, and connected across all of your systems by building in relationships and calculated fields, chord ai’s models act like a semantic layer, making the data more intelligible both to human users and ai agents this helps facilitate clearer query writing, more robust automation, and more meaningful insight generation key features remember to click the deploy changes button on the modeling page to sync any changes or edits you make within chord ai modeling ! this applies to any changes you make to the features below models, relationships, and calculated fields models models on chord ai combine schema definitions, metadata, relationships, calculated fields, and semantic logic derived from tables in your data source essentially, models help users and chord ai understand how your data is interrelated to create a model , follow the instructions below navigate to models on chord ai click “ add model ” select a table from your connected data source and click “ next ” select the columns you want to add to the model you selected, then click " > " to include them note selecting the primary key of your table is available and optional save your model relationships chord ai relationships define how data tables and data models connect, enabling more efficient querying and analysis of linked data to add a relationship to a model, follow the instructions below navigate to models on chord ai navigate to the specific model you wish to add a relationship select the actions menu icon click “ add relationship ” define your relationship properties from select the left side table and column of this relationship to select the right side table and column of this relationship type select the type of relationship (many to one, one to many, one to one) save your relationship additionally, note the following relationships cannot be created on chord ai self referential relationships a self referential relationship is when a column within a model points back to the same model chord ai currently does not support creating such self referential relationships, where a column in a model references the same model this limitation prevents the direct modeling of hierarchical or recursive data structures within a single model example in an employee model, the reportsto column may store the id of another employee , indicating the reporting structure within the organization duplicate relationships chord ai does not permit the creation of duplicate relationships between the same pair of models using identical columns this restriction ensures that each relationship remains unique, avoiding redundancy and potential confusion in model associations example if a relationship already exists between the order and customer models using the customerid field, you cannot create another relationship using the same customerid field between these models calculated fields chord ai’s calculated fields lets you create custom columns by applying expressions (similar to formulas in excel or in google sheets) calculated fields work like spreadsheet formulas for models, with a simple create/edit/delete flow follow the steps below to add a calculated field to a model you can also view the list of available expressions navigate to models on chord ai navigate to the specific model you wish to add a calculated field select the actions menu icon click “ add calculated field ” name your calculated field select the expression you’d like to apply save your calculated field your calculated field will appear on the model once saved additionally, if you pick a table that has relationships with other tables, you can "chain" them to drill into the other tables' fields within the calculated field modal to delete a pre existing calculated field , navigate to the relevant model with the calculated field you wish to delete click the ✏️ icon and click “ delete ” additional guidelines chord ai will now allow users to apply an aggregation (like max ) directly to a source model field this is due to the following source models usually contain “raw” or “dimension” data (like customer id , created at , product name ) aggregations ( max , sum , avg , count distinct , etc) are only meaningful when applied to fact like numeric measures or after you’ve joined into the right model/table context if chord ai allowed aggregation on arbitrary source fields, the results would often be misleading or undefined (e g max(customer id) isn’t a useful metric) chord ai errs on the side of protecting users from invalid queries alternate ways of applying aggregations would be to pick a measure field in a model that supports aggregation; or use relationships (joins) to hop into the next model where that aggregation makes sense