Every question, answered straight.

Every answer is built from two sources: your store's own policies and macros, and the verified customer's real Shopify order. Order lookups are bound to the customer's verified email, so nobody can ask about someone else's order. If those sources don't cover the question, Suvenna does not improvise. It abstains and escalates to your team. It never invents a tracking number, a discount, or a policy.

It says so and hands the conversation to your team with full context: the customer's message, the order, and what the AI checked. Abstaining is a feature, not a failure. An AI that guesses about refunds is a liability.

Email and storefront chat. Email connects through Gmail OAuth, Outlook OAuth, or a simple forwarding address, so your existing support address keeps working. Chat runs as a widget on your storefront. Voice, SMS, and social channels are not supported yet, and we won't claim them until they are.

Yes. The widget loads through Shopify's App Proxy, so it works with any Shopify theme and no code edits. When a logged-in shopper opens it, their identity is verified through Shopify's own HMAC signature, which is what lets the AI safely discuss their orders. Signed-out visitors still get policy answers.

Yes. Email attachments come through with the conversation, so a customer can send a photo of a damaged item and your team sees it right in the thread.

English first, honestly. Suvenna is built and tested for English-language support today, and we don't claim multi-language support until we can stand behind it. If most of your volume is in another language, tell us when you join the waitlist so we can be straight with you about fit.

Only if you turn that on, and only inside your rules. The customer must explicitly confirm before any money moves. You set a refund ceiling for auto-approval, and anything above it queues for one-click approval by your team. Out of the box, live actions are off entirely.

The order itself. At execution time, every refund is capped by the order's actual refundable amount in Shopify, so a stale conversation or a confused request can never over-refund. Your auto-approval ceiling applies on top of that, and the confirm-before-money gate is enforced in the engine itself, not in a setting someone can switch off.

Two layers. The customer explicitly confirms the action in the conversation, and risky actions then queue for one-click approval by a human on your team. Owner, admin, and agent roles let you control who holds that responsibility.

Yes, and it's append-only. Every money action and every consequential change lands in the audit log, with who or what initiated it and when it executed. Nothing in the log can be edited after the fact.

New stores start with every live Shopify action disabled. Suvenna can answer questions and draft replies from day one, but it cannot move money or change orders until you enable each action for your store. You flip those switches when you're ready, not us.

One customer thread your desk handles in a billing month, however many messages it takes to resolve. It counts once whether the AI resolves it, a human does, or both. There is no separate charge when the AI resolves something. That is the per-resolution fee we refuse to have.

Caps are soft. We never cut a conversation off mid-thread, and your customers never see a difference. If you're consistently over, we'll talk with you about the right plan instead of surprising you with overage fees.

Pay annually and get 2 months free, so a year costs the price of 10 months. Monthly billing stays available on every plan.

Stores that join through the waitlist and onboard in an early wave get 20% off their first year of Suvenna, with that pricing locked in for the full 12 months.

If Suvenna doesn't independently resolve at least 30% of your conversations in your first 60 days, we credit back what you paid. Independently resolved means the AI handled the conversation end to end, without a human stepping in. The numbers come from the same reports dashboard you see every day, so there's no private scoreboard.

Because a support desk migration deserves attention. Each wave is small enough that every store gets a clean rollout: policies loaded, channels connected, guardrails configured, and a human reviewing the AI's early work. We would rather grow in steps than onboard badly.

It depends on your position in the list and the size of the next wave, so we won't promise a date we can't keep. The list is ordered, we open waves as fast as clean onboarding allows, and you'll get real notice before your wave opens.

20% off your first year with that pricing locked in, a front-of-line position for onboarding, and the shortest path to our team. Feedback from the early waves directly shapes what we build next.

No. Joining takes an email address, nothing else. You add a payment method only when your wave opens and you decide to onboard.