video1.0<iframe src="https://www.loom.com/embed/82bbc6b267f24f98a32be7ea332bf759" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/82bbc6b267f24f98a32be7ea332bf759-0bcf8b981809aa90.gif393.582Hi, I am Prasant Gupta, and I built a simple ML based similarity system for KBuilder. I convert incoming prompts and existing inputs into vector embeddings using a pre trained Sentence Transformer model, then compute cosine similarity to find the best match. This enables semantic matching even when wording differs, like cost and pricing. I walked through the code flow, from loading the model and generating embeddings to selecting the highest scoring input and returning matches with similarity scores. I did not request any specific action from viewers.