Cdiscount is a French online retailer with a broad product catalog spanning electronics, home goods, and more, and its chat connection lets an agent search that catalog from a plain description instead of a keyword search that returns hundreds of loosely related results. Describe what's needed, a specific type of blender, a laptop for a particular use case, and the agent returns matching products pulled from Cdiscount's actual listings, then compares the strongest options against each other on price and features. Because the comparison happens in seconds within the same conversation, someone shopping across several categories for a household move or a gift list can evaluate each item without opening the site itself for every single search, jumping from one category to the next without losing track of what's already been decided. It works as well for a single specific item as it does for a broader browse across a category when the person isn't sure exactly what they want yet. A vague starting point, something like a gift for someone who likes cooking, still gives the agent enough to search on.
| Search catalog | Query Cdiscount's product catalog from a plain-language description of what's needed, without requiring an exact product name. |
| Compare products | Line up matching products against each other on price and features. |
| Filter by category | Narrow a search to a specific product category, such as electronics or home goods. |
| Check price range | Filter results to fit within a stated budget. |
| Evaluate options | Summarize the tradeoffs between a shortlist of matching products before a final choice is made. |
| Search across categories | Run several category-specific searches back to back within one shopping session. |
Someone moving into their first apartment gives the agent a list of things they need: a bed frame, a small dining table, and a vacuum, all on a tight budget with a move-in date two weeks out. The agent searches the Cdiscount catalog for each category, compares the top options on price and reviews, and returns one recommendation per item, noting where a slightly higher price buys noticeably better reviews and flagging one bed frame with a longer than usual delivery estimate. The person places all three orders in the same sitting instead of researching each separately across several evenings, and everything is scheduled to arrive before the move.
A student needs a laptop that can handle video editing without exceeding a set budget set by a parent helping cover the cost. The agent searches Cdiscount's electronics catalog, filters for the processor and memory specifications that actually matter for video work rather than general browsing specs, and compares three matching models on price and specs. It flags that one model, despite costing slightly more, has noticeably better cooling for sustained rendering tasks and a larger screen better suited to editing timelines, leaving the student with a clear pick instead of a long list of similar-looking listings to sort through alone.
No, describing what's needed in plain terms, the use case, the budget, the rough category, is enough for the agent to search Cdiscount's catalog and return relevant matches worth a closer look.
Yes, separate searches and comparisons across categories can happen back to back in the same conversation, which suits shopping for several unrelated things at once, like furnishing a room and buying a gift in the same session.
Matching products are compared within seconds of the search, well within a normal back-and-forth conversation, so there's no meaningful wait between asking a question and getting a usable shortlist back.
Yes, a price range can be applied so returned products fit within it, and that range can be adjusted mid-conversation if it turns out to be too tight or too loose for what's actually available.
It finds and compares products; completing checkout happens through Cdiscount directly, using whatever account and payment method is already set up there, so nothing is purchased without a final confirmation.
The search can be refined with additional detail, a different price range, a missing feature, until the results better match what's actually needed rather than settling for the first pass. Most searches only take a round or two of refinement before landing on a product actually worth ordering.