Use case / Customer service

Support that answers at 11pm
and knows what they bought.

One agent on support@, the website chat, your number and your DMs. It answers from your help docs and your own Q&A, reads the order before it replies, and hands the thread to you the moment a question is not one it should answer.

  • Email
  • Text messages
  • Website chat
  • Social DMs
  • Phone calls

7-day free trial · $25 a month · Cancel anytime

The same twenty questions, arriving on five channels

Most support is not hard. It is repetitive, it arrives at the wrong times, and it is scattered.

01

Where is my order, what are your hours, how do I return it

A handful of questions make up most of the queue, and each one is answered by hand, from memory, slightly differently every time.

02

The queue builds overnight

A message at 9pm waits until the morning. By then there are forty of them, and the person who sent the first one has already left a review.

03

Five inboxes, one customer

Email, the chat widget, Instagram, WhatsApp and the phone are each somebody's job, or nobody's. The customer who tried three of them is three tickets.

04

The answer is in a document nobody opens

The returns policy, the setup guide and the FAQ exist. They are just not in front of whoever is typing the reply.

What the agent does with a support question

It answers what it can from your material, reads the customer first, and knows what to hand over.

01 / Answers

From your documents and your answers

Upload the help docs, the policies and the setup guides, add Q&A pairs for the questions you get, and set your hours and prices as data points. Every reply comes from that and links only to pages you allowed.

02 / Reads

The customer before it replies

With the store connected it sees what they bought, what is still open and what they are worth. "Where is my order?" is answered about their order, not in general.

03 / Hands over

When a person should answer

A refund, a complaint, anything you named in its instructions: the thread lands in your inbox and the agent stands down on it. Reply, and hand it back when you are done.

Every channel in one list, with a reply box

The inbox is where every conversation the agent is having ends up: email, texts, call transcripts, website chat and social DMs, newest first, each row tagged with the channel it arrived on. Filter it to one channel, one number or one agent.

Reply in any thread yourself and the agent stands down on that conversation, so it never talks over you. Hand it back when you are finished.

The Inbox as one list: web chat, texts and calls together, a call marked no answer, a completed call that booked a roof inspection, texts confirming a Thursday appointment and asking the price, each row tagged with the channel it arrived on.

Support that stays consistent

The same answer, on every channel, every time, and a record of all of it.

Consistency

One knowledge base, every channel

The answer to "what is your returns window?" is the same in an email, a DM and a chat, because it is the same agent reading the same document.

Memory

One customer, one history

The chat cookie, the email address and the phone number are merged into one contact. Nobody has to explain their problem twice.

Record

Every reply stored with its source

What the agent said, and the material it said it from, kept against the contact. When a reply is wrong, you can see why and fix the source.

Isolation

One agent per brand

Run several brands or clients on one account. A knowledge base belongs to one agent, and content from one is never used to answer for another.

What you connect

Point it at the channels your customers already use. It answers from your own support address and your own accounts, so nothing about the customer's side changes.

Connect the store and it can talk about the order in front of it.

Mailboxyour support@ address
Websitea chat widget in your colour, with your greeting
SocialInstagram, Messenger, WhatsApp, Telegram and more
Numbertexts, and calls answered on Twilio
StoreShopify, WooCommerce, Stripe, Wix, Squarespace
KnowledgePDF and Word documents, Q&A pairs, data points

Before you sign up

What does it do with a question it cannot answer?

What you told it to. Its instructions say when to stop and fetch you, and a thread it hands over lands in the inbox with the agent standing down on it. It does not guess from outside the material you gave it.

Can it process a refund or change an order?

It can read the order from a connected store and tell the customer what it sees. Changing the order is a person's job: write that rule in its instructions and those threads come to you.

Can customers tell it is an AI?

That is up to you. The agent has a name, an image and a greeting you write, and its instructions decide how it introduces itself. Many businesses have it say so plainly.

Does it work with our existing help desk?

It works the channels directly: the mailbox, the widget, the number, the DMs. Conversations live in its own inbox, and with a connected CRM it writes notes and outcomes back against the contact.

How do we keep the answers up to date?

Edit the knowledge base. Replace the document, change the Q&A pair, update the data point, and the next reply uses it. There is no retraining step.

The queue is empty in the morning.

Upload the help docs, connect support@ and the chat widget, and read what it answered overnight.

7-day free trial · Cancel anytime