from Hacker News

Show HN: Human Chatbot

by txt-me on 8/3/22, 2:23 PM with 8 comments

We are a small team bootstrapping a human chatbot. It is like a chatbot, but real people answer the chats. You install our widget on a website, we connect our team and immediately start answering, 24/7.

How does it work? It is cloud customer service. The operators are shared between the websites, hence lowering the price.

What happens if the operator doesn't know how to answer? We collect the visitor's contacts and create a ticket. You then reply to the ticket, and the system sends an email to the customer.

Would love feedback from HN!

  • by jaclaz on 8/3/22, 5:52 PM

    I am clearly missing something.

    How would the agent(s) be capable of answering customers' questions?

    Besides the "ordinary" and "very basic" ones like (say):

    "Is the shop open next Sunday?"

    "I am sorry, no, the shop is open monday to friday from 9:00 to 18:00."

    And which kind of access should/would these operators have?

    I mean (one of your examples):

    "I am trying to pay, but my card is getting declined"

    "I am sorry, let me check it with the bank"

    What does (or can) the operator do, like:

    1) actually check with the bank

    2) open a ticket (or send a mail) about the issue

    Only to show how old I am, once upon a time customer support personnel (telephone) was "in-house" (employees that knew about the company operations and could actually solve problems in a timely manner).

    Then it was externalized to "call-centers", where operators were taught a small subset of the needed knowledge[1] and given very little power to resolve anything.

    Then it was (often) moved abroad and further restricted in what they could do, but still there was the need to teach lots of things[2] to the operators.

    If your operators assist for multiple companies, they will need to know quite a bit about each company, wouldn't they?

    How do you plan to transfer the needed knowledge to them?

    [1] AFAIK by means of (usually) quickly written "procedure manuals"

    [2] I believe by means of FAQ and FGA collections

  • by astonfred on 8/3/22, 4:58 PM

    Cool idea. $1 per hour makes it look very affordable (which it is, in a way) while generating a healthy $720 of potential MRR. Smartly priced. Can you share some numbers about the size of the current user base?
  • by IceMetalPunk on 8/3/22, 7:07 PM

    So... by "human chatbot" what you mean is "outsourced call center without phones"?