AI pre-sale tech support for electronics stores.
heysteff is an AI platform for customer support and sales across WhatsApp, Instagram, Messenger, Gmail and Shopify; Steff is the AI agent that runs it. In electronics, most pre-purchase questions are technical — compatibility, specs, differences between models — and answering them well or badly decides whether the sale closes or walks to another store with faster support.
Compatibility and specs from your knowledge base
You load manuals, spec sheets and FAQs into the agent's trainable knowledge base. When a customer asks whether a charger is compatible with their model, or how much RAM a machine supports, Steff answers with that specific information — not with a generic or made-up reply. If the question needs a data point that isn't loaded, the agent says so and escalates, instead of risking an incorrect technical answer that ends in a return.
This matters especially in electronics because a wrong technical answer costs more than a lost sale: a customer who buys an incompatible accessory because of a bad answer ends up with a return, a complaint and a negative review. The "don't make it up, escalate when you don't know" rule protects both the customer and the store's margin.
A model comparator with real catalog data
For "which one should I get, model A or model B?", the HTTP API and webhook tool builder lets you connect your real catalog — price, specs, stock — so the agent uses that data in the answer through a tool, instead of reciting a stale static spec sheet. The comparison comes out with the current state of inventory: if a model is about to sell out, Steff can say so in the same message.
The difference from a static comparison page is that the conversation adapts to what actually matters to that specific customer: if they asked about battery life, the answer emphasizes that; if they asked about price, it emphasizes that instead. It's the same catalog information, ordered around the real question.
Warranty and RMA with rule-based escalation
Warranty claims follow the process you define: the agent confirms the problem, walks through the first diagnostic steps if your policy includes them, and assembles the RMA request. When the case warrants human intervention — a device with a complex fault, a policy exception — Steff escalates via rules and keywords with the full context of the conversation, and pauses itself so it doesn't step on the reply of the person taking over.
If the customer sends a photo of the damaged device or of an on-screen error message, Steff reads it thanks to vision and attaches it to the ticket, so the person picking up the case already has the visual evidence without having to ask for it again.
Order tracking without leaving the chat
Order status — shipped, in transit, delivered — is checked right from the conversation, connected to your store or your shipping system. It's one of the most repetitive queries in any ecommerce, and taking it out of your team's queue frees up time for the technical questions that genuinely need human judgment.
Where copilot mode comes in
Not everything gets automated end to end, nor should it: there are electronics questions the technical team itself wants to answer personally, or where the customer asks to talk to someone. With copilot mode, the conversation can be taken over in one click without losing the history the agent already built — specs consulted, comparison shown, warranty ticket in progress — and the bot pauses itself so it doesn't interfere while the person replies.
◆ the star flow
The model comparator with real data is the one that moves the sale most: it answers in seconds a doubt that, without up-to-date data, an undecided customer usually resolves by buying somewhere else.
Related
◆ next step
Get a demo and connect your real technical catalog to the agent.