Takes store actions
Finds the order from the caller's number or the order number, checks whether it has shipped, opens an address-change or return request in your team's inbox and books callbacks.
Why Zitena
Not a call-center bot. Your store's hands on the phone.
Generic voice bots answer questions and leave the work to your team. We built for e-commerce from the start: the agent finds the order, opens the request, calls about the cart and shows everything it did in the console.
Comparison
Generic voice bots vs Zitena
| Topic | Generic voice bots | Zitena |
|---|---|---|
| Takes store actions | Mostly answers FAQs and leaves the real work to your team. | Finds the order, reads tracking, opens address-change and return requests, books callbacks. |
| E-commerce native | General purpose; you have to describe orders, shipping and carts yourself. | Orders, tracking, carts, return policy and catalog are modelled from day one, with setup per platform. |
| Purpose-driven outbound with consent | Upload a list and dial. Checks are often left to you. | CSV list campaigns are supported, but every row is still called with a purpose and a subject; İYS consent, calling hours, the do-not-call list, quotas and the repeat guard are checked before dialing, for campaigns too. |
| Live console | A recording and an after-call summary. | Live transcript, response time on every turn, the tools the agent used and the requests it opened. |
| Topic analytics | Manual tagging or a generic sentiment score. | Every call is tagged by topic automatically, with period trends, "solved by the assistant" rate and a manual correction loop. |
| Sounds human | Speaks in strict turns; gets confused when interrupted or when you pause. | Natural barge-in, waits for unfinished sentences, and a ready phrase library: acknowledgements and fillers are generated in advance in the agent's own voice and quality-checked. |
| Personal data | Masking and encryption are worth asking about. | Numbers are encrypted at rest and masked in the console by default; for post-call analysis, phone numbers, emails, addresses and long numbers are masked before a transcript goes to a language model. |
The comparison describes the general-purpose voice bot approach, not any specific product.
Concrete differences
Not claims. Things you can see on screen.
Built for e-commerce
Orders, tracking, carts, return policy and catalog are modelled from day one, with guided setup for Shopify, WooCommerce and Turkish platforms.
Purpose-driven, consent-checked outbound
Whoever starts the call writes its subject and the agent says it first. İYS consent, calling hours, do-not-call list, repeat guard and quotas are checked before dialing.
Response time on every turn
The live transcript shows each agent turn's response time, the tool it called and the result. You don't guess at slowness, you see it.
Topic analytics
Every call is tagged by topic automatically, with period trends, the share solved by the agent and a manual correction loop on one screen.
Encrypted personal data
Numbers are encrypted at rest, matched with a keyed HMAC and masked in the console by default. For post-call analysis, phone numbers, emails, addresses and long numbers are masked before a transcript goes to a language model.
Sounds human
Sounding natural on the phone is in the details.
Even on a short call, customers interrupt, pause and give the address in pieces. The agent handles that:
- Natural barge-inWhen the customer cuts in, the agent stops and listens.
- Waits for unfinished sentencesIt doesn't jump in while a sentence is still going.
- A ready phrase libraryAcknowledgements and fillers like "sure" or "let me check" are generated in advance in the agent's own voice and quality-checked.
- Reads addresses and numbers naturallyOrder numbers, tracking codes and addresses are repeated in easy-to-follow chunks.
- Warns before a long call endsAs the call approaches its time limit, the agent tells the customer a minute ahead, so it never just cuts off.
Sample call — voiced re-creation
Not a real customer call: a voiced re-creation of a scenario that mirrors the live product flow. The live agent speaks Turkish; we tailor it to other languages on request.
Agent Hi, thanks for calling Erusoft, this is Deniz. How can I help?
Customer Hi, yeah. I ordered some headphones last week, and I'm just wondering where they are.
Agent Sure, I can check that. What's your order number?
Customer Uh, it's one oh four two two... sorry, no. One oh four two one one.
Agent One, zero, four, two, one, one. One moment, let me check.
Agent Okay, found it. Your order is being packed, and it hasn't shipped yet. It's due to go out tomorrow morning, and once it does, you'll get a text with the tracking—
Customer Oh, sorry, that's actually why I'm calling. If it hasn't shipped, can I change the delivery address?
Agent Yes, since it hasn't shipped, we can still change it. What's the new address?
Customer It's Flat 3, 14 Kestrel Road, Hackney, London, E8 4QX.
Agent Thanks. So that's Flat 3, 14 Kestrel Road,
Customer Mm-hm.
Agent Hackney, London, E8 4QX. Is that right?
Customer Yep, that's it.
Agent Great. I've created the address change request, so your order will go to the new address. Would you like a text confirmation to the number on your account?
Customer Yes, please.
Agent Done, that's sent. Is there anything else I can help with?
Customer No, that's everything. Thanks a lot.
Agent You're welcome. Have a great day.
Proof
We don't make up social proof.
You won't find customer logos, quotes or "save X%" figures here. Instead, things you can judge for yourself:
- In a pilot with a Turkish electronics retailer.We don't share their name without their permission.
- A sample call recordingA voiced re-creation of a scenario that mirrors the live flow, in the player above. The live agent speaks Turkish; the English version is a re-creation.
- Real screenshotsEvery screen on this site comes from the real console, in a demo workspace with fictional data.
- A live call in the demoIn the demo we call the agent together, using a scenario from your own store.
Evaluation checklist
Questions to ask when evaluating any voice agent.
Useful whether you talk to us or anyone else. In the demo we answer every one of them on screen.
- Can the agent actually find an order, or does it only answer FAQs?
- Does it open a request for your team for an address change or a return? Where does it show up?
- What happens to a transfer when nobody answers: does the call drop, or is a callback booked?
- Can you see the response time of every turn, or only an average?
- Are commercial-communication consent, calling hours and a do-not-call list checked before outbound calls?
- How are phone numbers stored, and are they masked in the console and logs?
- At which steps does personal data reach a language model, where is it masked, and in which country is it processed?
- How is the vendor's support access to your account limited, and is it recorded?
- Can you version settings and roll back?
- Can you keep your own number and SIP provider?
Hear it on a call with your own store's data.
In thirty minutes we walk through the console, place a live call and work out which of your calls the agent can take.
or write to [email protected]