[wongarsu]

I once wrote a chat bot for a printer (think big photocopier with network printer functionality operated by student body of ~10 students and used by ~2000 students).

It had simple commands for showing the state of consumables or monitoring budget spent (calculated based on spent consumables). But the real value came from the chat bot requesting actions ("yellow toner at 12%, verify replacement is in storage", "A3 paper is running low, refill suggested", "heavy A4 paper empty, replace now", etc).

Of course all of this could have been it's own app with push notifications. But getting people to install that would have been much harder than having them use a chat bot in a messenger they already have. Using a chat bot also made it available everywhere, on Android, iOS, Windows, the Linux distribution of the week, if you desire even your watch or fridge, with synchronised notifications. And all that with a bit of code running on some server, none of the head aches of front-end development.

[rena-anju]

Thank you for sharing. I have always considered chat bots to be passive, but this opened my mind. Maybe I am narrow minded...

[wongarsu]

I think of chatbots as asking the question "what if we could put an AGI (artificial general intelligence) in here" and programming the closest approximation for the task. Sometimes that's a natural language interface, and that's what most people focus on (lots of money in customer support and sales automation). But if our office coffee machine had an AGI I wouldn't care about natural language comprehension, I want to be able to tell it to have coffee ready when I come to the office early and I want it to convince humans to clean it.