The Nexa AI Assistant for Shoppers and Sellers

Nexa runs one intelligence layer with two faces. For shoppers it is a shopping assistant that searches real inventory and helps build a cart. For sellers it is a business assistant that reads their own orders, revenue, and inventory and explains what is happening. Both are grounded in live platform data and both operate strictly inside the permissions of the person using them.

Key takeaways

  • The assistant reads live catalog and account data, not a stale snapshot.
  • Shoppers get product suggestions; every cart addition requires confirmation.
  • Sellers get revenue, inventory, and growth insight in plain language.
  • Permissions are enforced server-side, so the assistant cannot cross accounts.

For shoppers: from intention to cart

The shopping assistant is available across the marketplace as a floating button, so it is reachable from a product page, a category, or the middle of a cart. It accepts natural language, remembers the thread of the conversation, and narrows as you add constraints.

  • Ask for a product by purpose, budget, recipient, or occasion rather than by keyword.
  • Compare two products with the tradeoffs stated plainly instead of as a spec table.
  • Check availability, sizes, and colors without leaving the conversation.
  • Get a suggested cart for a scenario — a dinner party, a gift set, a starter kit.
  • Confirm each addition yourself; nothing enters the cart silently.

Grounding is what separates this from a generic chatbot. Every product the assistant names exists in the catalog with the price and stock it reported at that moment. If something is out of stock, it says so rather than recommending it anyway.

For sellers: answers instead of exports

The seller-side assistant reads the seller's own business records — orders, revenue, refunds, processing cost, inventory, and payouts — and answers questions about them. This is the practical payoff of running payments and commerce on the same data layer: margin is a fact the system already knows.

  1. Understand performanceWhich products drove revenue this month, how that compares with last month, and which categories are trending in the wrong direction.
  2. Find the real marginRevenue after refunds and processing cost, per product, so pricing decisions are made against net rather than gross.
  3. Manage inventoryWhat is selling faster than expected, what has not moved, and what should be reordered before it runs out.
  4. Spot growth opportunitiesBundle candidates, pricing adjustments, and auction timing suggestions drawn from observed demand rather than generic advice.
  5. Draft the workProduct titles and descriptions, category assignments, promotional copy, and storefront sections generated and then edited by you.

Deep insight views

Beyond conversation, key dashboard tiles open into deeper AI-written analysis: what changed, why it likely changed, and what could be done next. The analysis cites the underlying numbers so a claim can be checked rather than taken on faith.

How permissions work

The assistant does not have privileged access. It issues queries as the authenticated user, and the same database-level access rules that protect the dashboard apply to those queries. A shopper's assistant sees public catalog data and that shopper's own orders. A seller's assistant sees that seller's store. A staff member's assistant sees only what their role permits. There is no path by which the assistant reveals another account's data, because the data layer itself refuses the request.

Where AI shows up elsewhere in the platform

  • Storefront building from a plain-language description of the brand.
  • Product import cleanup: titles, descriptions, categories, and variant photo matching.
  • Auction assistance: research, condition notes, and starting price guidance.
  • Catering and custom-order builders for food and hospitality sellers.
  • Automated operational passes that flag reorders and tune listing parameters overnight.

Limits worth knowing

The assistant is strongest when a question can be answered from platform data. It is not a tax advisor, a lawyer, or a substitute for professional financial advice, and it does not have visibility into systems outside Nexa unless those systems have been connected. When it does not know, the right answer is that it does not know — and the underlying data is always available to check.

Frequently asked questions

Short, direct answers to the questions people most often ask about this topic.

Continue through the Nexa OFS Knowledge Center.

Core pages across the Nexa ecosystem.

Looking for something else? Browse the full Nexa AI Knowledge Center.