Customer support agent. Delivery, returns, availability: the questions that fill your inbox get answered on sight. Anything that needs judgment goes to a person, with the context attached.
Aura Skincare
online
Answer every customer in seconds, day or night. The agent works from your knowledge, hands off with a reason, and never guesses.
Aura Skincare
online
Aura Skincare
online
Aura Skincare
online
Appointment booking
Books time with a native WhatsApp form, right inside the chat.
FAQ from your docs
Answers from the pages and files you provide. Nothing invented.
Order status
Connect your store’s API and it answers “where is my order” itself.
Custom
Your own instructions, tools, and knowledge, scoped to one job.
Agents also run as a step inside your WhatsApp automations, answering mid-flow before a human takes over. And if you run a store, order status is the natural place for D2C brands to start: “where is my order”, answered from your own systems. Every plan includes a monthly quota of AI actions, and pricing scales by volume, not seats.
Crawl your site
Point it at your pages. Policies, shipping, product details.
Upload files
Price lists, brochures, and documents become answers.
Write articles
Short internal notes for the things only you know.
When the answer is not in your knowledge, the agent says so and hands the conversation to your shared team inbox, with the reason attached. It never fills gaps with the internet’s opinion of your refund policy.
Hands off with a reason
Escalates to your team instead of guessing, with a one-line why.
Quiet when humans reply
Auto-mutes the moment a teammate types, no talking over.
Needs-human filter
Escalated chats surface in one inbox view, nothing slips.
Per-conversation toggle
Turn the AI off for any chat, and back on when you are done.
Converse with your draft agent in a playground where simulated sends never reach WhatsApp, and inspect every reply: which knowledge it retrieved, which tools it called. The agent can only go live once the checklist passes.
It is the step past a chatbot. A rule-based chatbot follows a fixed script. A Wapflow AI agent reads the question, answers from your own knowledge, uses tools like order lookup, and hands off to a person when a person should decide.
It replies in the language the customer writes in, including Hindi and Hinglish. A customer who writes in Hinglish gets Hinglish back, not formal English. The answers still come only from the knowledge you provided: your site, your files, your notes. The language adapts; the facts and the prices do not.
Yes. Every agent starts in a sandbox playground where simulated conversations never reach WhatsApp, and it can only go live after a checklist passes: escalation path set, tools tested, at least one sandbox conversation, quota available. In the playground you can inspect every reply, which knowledge it retrieved and which tools it called, so you see why it answered before it answers for real.
The agent pauses silently and conversations route to your team inbox, with a notice to you and threshold warnings before it happens. Customers never see an error; they just get a human. Every plan includes a monthly AI-action quota, from 100 actions on the smallest plan to 25,000 on the largest, so you can size the plan to your volume.
Yes, through tools. Connect your Shopify, WooCommerce, or any REST API as a tool, and the Order Status preset answers from your own systems. Every tool must pass a test run before the agent can use it live. The preset looks the order up and replies with real tracking, not a canned line. There is no native Shopify app yet; the connection is your store’s API.
Each agent is scoped to one task, never a general-purpose persona. It discloses that it is AI, is hardened against prompt injection, and never invents prices, policies, or availability. When a question falls outside its knowledge, it says so and hands the chat to your team instead of improvising. And you can turn it off for any single conversation with one toggle.
Early access is rolling out to the waitlist in order.