Will the agent actually save a subscription cancel, or just deflect it?
Save, not deflect. The agent reads the customer's Recharge history, understands the likely reason for the cancel (frequency, life event, product fit, price), and offers exactly one regulator-compliant alternative — a pause, a swap, a downgrade, or a skip — that fits this customer's pattern. If the customer says no, the cancellation processes immediately. This is the FTC click-to-cancel rule, enforced at the workflow level. Our cohort save rates land between 38% and 71%, depending on category.
How does this fit alongside Gorgias or Zendesk — does my team see the same view?
Yes. Repliant writes every conversation back to Gorgias or Zendesk as a ticket with the thread attached, the systems calls listed, and the disposition tagged the way your team already tags them. Your CX team's view does not change. They see the agent's resolved tickets in the same queue they always have, and they see the cases the agent escalated with full context one click away. The two systems coexist — we are not asking you to migrate off your desk.
Can it process a Loop or Returnly return without a CX agent touching it?
Yes. End-to-end: eligibility check against your policy, photo capture if you require it, replacement-or-refund decision, return-label generation via Loop / Happy Returns / Returnly, refund issuance through Stripe or Shopify Payments. Your CX team sees the resolved ticket in Gorgias the next day. The cases that genuinely need a human — disputed reason codes, final-sale exceptions, lost-in-transit edge cases — escalate with full context. Roughly 4% of returns in our cohort end up at a human; the other 96% close without anyone typing.
How do you handle TCPA on outbound SMS — and can it co-exist with Attentive or Postscript?
Outbound SMS sends only to numbers with a recorded TCPA opt-in tied to your store, time-stamped and exportable. STOP / HELP / SHAFT keywords, quiet hours (8am–9pm local), and per-state restrictions are enforced at the channel layer. Repliant runs alongside Attentive or Postscript without competing for the consent record — we read from your existing CDP or directly from Shopify customer profiles to verify consent before send.
Can it handle BFCM — our volume goes 14× a normal day?
Yes. Capacity scales horizontally with the surge, with no warm-up period required. The interesting question is not whether the agent can scale — that part is uninteresting — but whether it stays composed at scale. We run a BFCM drill with every customer in their first deployment: synthetic traffic at 20× normal, on every workflow. Brands that miss this step in their first peak find out the expensive way. Brands that don't, don't.
Does the agent actually speak in our brand voice across email, SMS, voice, and WhatsApp?
Voice is calibrated from a sample of 50 tickets your CX manager approves — including the channel-appropriate register (email is more formal, WhatsApp is briefer, voice has its own rhythm). Each channel inherits the brand's tone but speaks the channel's grammar. Brand-voice changes ship through your existing review process; your CX manager signs off on every meaningful update. We do not invent catchphrases.
How do you avoid hallucinated product info — sizes, ingredients, ship dates?
The agent quotes only what it has retrieved — prices and variants from your Shopify catalog, ingredients from your PDP fields or a structured ingredient table, ship dates from ShipStation. Anything outside the retrieved scope triggers a "let me check with the team" handoff. The agent cannot make up a fragrance-free variant if you do not stock one.
What's pricing for a brand at our volume?
Consumption-based, with a monthly floor. The brands in this category typically replace the cost of two to four full-time CX hires with a Repliant contract that's a fraction of the loaded cost. We quote per conversation, not per seat — your CX team is not the cost line we are trying to reduce. The cost we are trying to reduce is the line your CFO calls "the support number we cannot explain."