System Design · Airtable · Hard
Imagine a no-code table product (similar to Airtable) where users store data in named fields and write expressions in the product's own formula language. Build an ML-driven assistant that interprets a plain-language instruction and produces a syntactically valid formula grounded in the current table's schema. You have access to both the table platform's API (read/write) and a hosted LLM API. The assistant may receive requests such as incrementing a numeric column, creating a…
Checking your access…