ProjectIntent Over Keywords: From Stitched Embeddings to a Production Fashion Agent · Module 14: The planner: intent before retrieval · part 16 of 26
Part 16 · Module 14: The planner: intent before retrieval

Module 14: The planner: intent before retrieval

23 min read·
About 50 minutes · Needs: Module 11 (StyleMe, the product problem), Module 12 (high-level design), Module 13 (sessions) · Code: part3-styleme/backend/src/styleme/agent/planner.py, agent/parsing.py, agent/prompts.py, agent/graph.py, originals/styleme/fashion-agent/agent/nodes/planner.py

Where this fits

WHERE THIS FITS Intent Over Keywords this module PART 1: STITCHED VIDEO RAG three models, two spaces, RRF PART 2: UNIFIED VIDEO RAG one model, one space PART 3: STYLEME FastAPI, LangGraph, React QUALITY AND OPERATIONS evals, Docker, Cloud Run Frames ffmpeg, one per 5 s CLIP ViT-B/32 frames + text, 512 d Whisper base speech to text MiniLM 64-word chunks, 384 d Two collections p1_visual, p1_text Rank fusion RRF, k = 60 Gemini answer from transcript text API + Streamlit upload, ask, sources Windows ffmpeg, video + audio Embedding 2 Gemini, one vector One collection p2_unified, no fusion Notes + answer Gemini, cited times React web chat, photo, chips nginx one URL, /api proxy FastAPI /api/chat, limits Sessions in memory, cap of 3 Planner intent before retrieval Implementer batched embed, search Generator stylist reply, ids Grounding check no LLM, drops invented Gemini flash-lite, embedding 2 Qdrant styleme_products Catalogue ingest image + title vectors Langfuse optional traces Test video 26 questions, 4 kinds Video eval Part 1 vs Part 2 Agent evals intent, grounding Guards uploads, limits, secrets Containers compose profiles Cloud Run StyleMe only CI/CD GitHub Actions, WIF Original code v1, v2, fashion-agent
This module (highlighted): the planner, the Gemini model it asks (amber, a model) and the session history it reads; blue is our own code.

Module 12 (high-level design) drew StyleMe as a graph of three nodes. This module opens the first node, the planner, the part that gives the course its name: before StyleMe searches anything, it decides what the shopper wants. The rest of Part 3 depends on this decision being made well, and safely when the model gets it wrong.

The rest of this project is yours to keep

This project is bought on its own, once, and stays readable afterwards, including the parts added to it later.

Or open everything with a plan