{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/c4c1e66e5b01458e80066be8e8618225\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/os/loom-media-content/loom/d5da1dd2-d5c3-40eb-b449-8c5948cf5227/c14cec01-c4a1-4f9e-9649-f4e6bc5efa81/sessions-thumbnails-c4c1e66e5b01458e80066be8e8618225-c41cab0737ece069.gif/binary?authToken=eyJraWQiOiJtaWNyb3MtZ3JvdXAvdGRwLW9zLWFyY2hldHlwZS9xNDk5YWRqaGoxZHE4aThhIiwiYWxnIjoiUlMyNTYifQ.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.SEy5_oW9kGll1iqvULH58yYB-Za37oVIdM_sf63pxDXh8Sjx_jL89B0V7SLTYJJBPFVQgHRLVHG42jKnoa5ETHrni0YkdXV5PVxcKpacJTEPIZWB54AWN9RR4huopk09hajkKrxswpotIet3VPvoWg0AEkIxwguHiH9tW2ZPc-wjBb9j2m4nlWPrrQDQN9HJLLowQ-jY4glpsanjYgU4vmalAYdS2WnBtFpKV92HjeS_veYG_vfPU1SPCar6sAmkJmj4l_45xfhtj0eYhFh6IsLJHuY4sS1-_4JqBGEDb49sG16PVRblXqj1o7G1dmkwwfutCUSWcG2bKGDTya5-pw&Expires=1795144178&Signature=A2INHhbtKZDDlzWTZXxG5jryoqTdsg3%7EInkNbQklG%7EjGTMxXaWc95%7E8krq2reZBgDH39BikgR%7EThNMP2e%7EcbkhSEnN8ojxTHcuVtC%7Eh3FgQ-leC89qwJ5h5QRu24GwyJ0Sl7AVx4qtmR8bBQOxAS2lRMmhIjmPuxzMGWYkGkwPD5bWyFFVEOzod-oO0YZFRx6v7h-5I1-5DH1-LIjHjpaL1HIqB9zAAsa1qAPVzlRqPmNKWEqEXLi7uDxMeAZH0Z4fxkC5Aa31251CxC4SLDQ5s5lWLToouCVbcWM7cFP5wWfl5%7EqJkNVAGbNkIyJVx9osjYW2qagd-FzT-lqI9Kyw__&Key-Pair-Id=KQOSYIR44AIC0","duration":148.524,"title":"Kala Fastpath: Natural Language to KALA Queries","description":"This Loom explains the Kala Fastpath pipeline for converting natural language into KALA query format. Vincenzo describes how a small language model generates an intermediate result that a compiler then translates into a correct KALA query, aiming to reduce ambiguity and excess aggregation (for example, questions like “who is funded” returning more data than requested). He also notes difficulties integrating with the Pioneer provider and states that the demo is fully controlled rather than using the real model. Finally, he mentions that the repo model’s results are not as good as Frontier models and ends by outlining his need to continue improving."}