Use case · For your business

Customer support at any hour

Ada reads each incoming message and answers the questions your team has already settled, learning from each reply you approve. When a refund is over your limit, a customer is upset, or a question is new, it comes to you with the background already pulled together.

I reply the way you would, and I check with you when I’m unsure.

Ada

At 3 a.m. on Saturday, a customer in Sydney gets a correct, friendly answer in four minutes. You read about it on Monday in the weekend summary.

A weekend at the front desk
  1. Sat 3:12 am

    A customer in Sydney can’t find their invoice. Ada tracks it down, sends it over, and explains where to find it next time.

  2. Sat 9:40 am

    A teammate asks about a wholesale discount. Ada finds last month’s decision and links it.

  3. Sat 2:05 pm

    Ada asks Rowan to check a shipping question on the carrier’s site. Rowan confirms the new cutoff time.

  4. Sun 6:30 pm

    “A customer wants $240 back on order 1187, which is over your limit. Do you want to approve it, or should I offer store credit?”

  5. Mon 8:00 am

    The weekend summary: 41 answered, 2 waiting on you, and 1 new question worth adding to the help page.

How it works

The pieces behind the story

Where it runs

Ada has a private computer, either in our cloud or on a spare Mac at the office, and reaches only the inbox and apps you connect.

Who does it

Ada covers the front line and remembers each answer you approve. Rowan helps with research.

What you control

You set refund and spending limits once, and anything above them waits for you to decide.

Your customers’ messages stay in your workspace, and we never sell them. You pay one predictable price, with no charge per person.

Set it up

Getting set up

  1. 1

    Give Ada an email address and connect Slack or Teams.

  2. 2

    Point Ada at your help pages and a few past answers.

  3. 3

    Set your limits and let Ada answer the next message.

Keep exploring

More use cases

Browse all use cases
Make a start

Start with one Agent and one real job

Set up a workspace, give an Agent something useful to do, and see how far it gets.