Open your own store chat at 11 PM and ask “where is my order?” Silence until morning? The Zanus chat answers in seconds, from the order record, in the shopper’s language — and the cart survives the night.

A shopper asks for 40 units of a SKU: the AI quotes from your real catalog prices and sends a payable link — no invented discount, no morning-after spreadsheet.

“I need to return these boots” starts an RMA with the order attached, quotes YOUR return policy word for word, and hands your CX lead a ticket instead of an angry email thread.

A weekly AI review of the returns log: which SKUs bounce, which supplier’s sizing runs small, which carrier damages boxes — drafted with the records cited, on your desk before the coffee.

The CRM molds to how retail actually talks: vendors, MOQs, lead times, payment terms. Ask “which suppliers can restock before Black Friday?” and get an answer with the contracts attached.

Every AI-drafted document — the price-list reconciliation, the supplier email, the promo brief — lands in Output awaiting review. Nothing ships without your decision.

PDF catalogs, supplier spreadsheets, scanned contracts, product photos, packaging artwork — the ingestion pipeline turns what lands in your inbox into queryable knowledge.

The same library that answers your storefront chat answers your staff: when the return window changes, you update one document and every answer everywhere updates with it.

The warehouse asks about carton dimensions; the buyer asks about supplier terms: each gets answers scoped to their role, enforced server-side.

Returns review, price-list reconciliation, promo brief, listing audit: work types molded to your operation, with default libraries and examples per type.

The AI’s workload is managed like the team’s: one-click views show what is urgent, late, today and tomorrow — nothing slips silently into next week.

Every work is a full session: reasoning streamed, sources cited, milestones tracked, linked to the supplier or the campaign it belongs to.

Return handling, price-match rules, damage claims, opening and closing checklists, marketplace policies: one library, current versions on top, superseded drafts outranked.

“Can I take this return without a receipt?” — answered in seconds with the policy attached, on the floor, on a phone, at Saturday peak.

Work types carry their default playbooks: the AI drafting a refund email reads from the current return policy — by construction, not by memory.

New floor staff learn your return rules, POS workflows, escalation paths and brand voice through guided training — ramp-up in days, ready before the rush.

“How do I process a marketplace return?” — asked to the assistant, answered from the playbook, without pulling the store manager off the floor at peak.

Private chats work on any device with a login — the how-to knowledge follows the staff member to the register, the stockroom or the pop-up location.

Return request arrives → the AI reads the order record and the return policy, drafts the RMA and the customer email, and queues both for your CX lead’s approval.

New review lands → the AI drafts a reply in your brand voice — grateful for the five stars, concrete about the two — and files it for approval before anything posts.

Every day at 7 AM: orders stuck, shipments late, payment flags — compiled from your records into one digest your e-commerce manager reads with the first coffee.

Bought-twice-lapsed-six-months, VIP-spend-tier, size-medium-boot-buyers: audiences built from your own customer categories and purchase history — no exporting lists to a third-party tool.

Announcements and offers written the way your customers actually read them — in your voice, in their language, grounded in the products you actually stock.

Opt-outs and communication preferences are enforced by the system, not by memory. Every send is on the record.

Orders, customers and products mapped field by field, with a dry-run preview before anything writes — push, pull or two-way, your choice per object.

Salesforce and HubSpot connectors sync contacts and deals push, pull or two-way — the wholesale side and the DTC side finally read from one brain.

A full API for whatever else the store runs: the 3PL portal, the marketplace feed, the POS export — data in, results out, under scoped keys.

The server redacts what a role cannot see — seasonal floor staff see the order status, not the customer’s purchase history or the supplier’s cost column.

A searchable log of every audited action: who exported which list, who changed the return policy, who viewed the wholesale price file, and when.

Five-level workgroups map stores, channels and departments; the on-premises option puts the whole system — catalog, customers, costs — inside your building.

Ask the chat: “where is my order?” or “do you have this in a size 42?” — watch it answer from a real catalog, in your language, without inventing a thing.

Ask: “how would you handle a return request?” — it walks the RMA workflow it would run in your tenant, grounded, cited, approval included.

The full demo loads a retail tenant with a sample catalog: order questions, returns, campaign drafts, supplier reconciliation — your store, working.