Find the right piece.
Every search. Every session.

Input: "Muhurtham set, traditional South Indian, 22K, budget ₹5 lakh"

What happens:
→ Extracts: Occasion=Muhurtham, Style=Traditional, Purity=22K Gold, Budget=₹5L
→ Filters catalogue by those parameters using TF-IDF scoring
→ Sorts by Desirability Score (5→1) as tiebreaker — same logic for walk-ins and video clients
→ Returns top 5 pieces with price in INR and USD, story, image URL, stock status
→ Works identically for a customer at the store counter and for a customer on a video call from California

Zero LLM. All retrieval and ranking is database logic. No AI tokens consumed.

Input: Associate taps any piece — the agent assembles the full set instantly.

What happens:
→ Reads the Set_Group value of the selected piece
→ Retrieves all pieces sharing that Set_Group
→ Aggregates: total gold weight, total INR, total USD, stock status per piece
→ Flags if any piece in the set is already Sold

In-Store use: Associate shows a necklace at the counter. Taps "View Full Set" on the tablet. The matching earrings and bangles appear instantly, with total set price.
Video / WhatsApp use: Before the session, associate sends a WhatsApp summary: full set, total price in INR and USD, availability across stores.

What it does: Finds pieces in your catalogue that no staff member has ever logged an interaction for — never shown in-store, never sent via WhatsApp, never included in a video session.

How "unseen" is defined: A piece is unseen if it has zero entries in the Interaction Log (the notes you write in the piece's detail view). The more consistently your team logs sessions, the more accurate this scan becomes.

Three scan modes:
→ All Customers: Any piece with Desirability Score ≥ 4 or Price ≥ ₹1L, never shown
→ India Walk-ins: High-value pieces (≥ ₹1L) not shown at in-store sessions
→ Overseas / Remote: High-desirability pieces (Score ≥ 4) not shown in video or WhatsApp sessions

Why it matters: This scan surfaces catalogue value sitting idle in your vault. It gives your team a morning briefing: "These 4 pieces worth ₹8.2 lakh have never been shown to anyone."

Problem it solves: You want to show specific pieces to a customer — via video call, WhatsApp, or email — without them seeing your internal notes, desirability scores, staff reactions, or admin flags.

How to use it:
1. Go to Smart Search. Enable "Select for Gallery" mode (button next to result count).
2. Click pieces you want — each highlights with a tick mark.
3. Click "Create Gallery Link" in the bar that appears at the bottom.
4. Enter the customer's name and a session label. Click Create.
5. A unique link is generated. Copy it and share directly with the customer via WhatsApp or email.

What the customer sees: A clean professional gallery page with piece name, photo, design story, price (INR and USD), and BIS HUID. Nothing else.

Security: The link expires after 7 days. Internal fields (score, flag, notes, reaction, staff name) are stripped before sending — they never appear in the customer-facing URL.

When Zithara CRM is connected: The system will pull upcoming occasion dates for all customers — weddings, anniversaries, Muhurtham dates — and automatically surface the most suitable pieces from the Image Bank 30 days before the event. A WhatsApp draft is prepared for the associate to review before sending.

Applies equally to India walk-in customers, overseas video clients, and WhatsApp-only customers.

When automated: Every Friday at 6 PM, a scheduled job reads Interaction Logs across all pieces and all stores, counts sessions per piece, identifies shortlisted-not-converted items, and sends a formatted digest to the store manager. Covers both in-store India sessions and overseas video/WhatsApp sessions in the same report.

Available now (manual): The Daily Intelligence Digest in the CRM / ERP tab generates this report from your current data. It enriches automatically as staff log more sessions.