Just curious if anyone is doing anything with chatbots yet?
What are you building on? Why did you choose that?
What service are you trying to provide or problem are you solving?
What is the high-level strategy of your chatbot?
What are the outcomes and goals of your chatbot?
Is this something you are considering? if so, why?
How do you think it could be valuable?
Would love to hear what people are doing/thinking on this subject.
I’ve been mulling around what could be done with a chatbot to bring value. It seems like basic information finding is low-hanging fruit, but not sure how valuable that really is, especially given the mediums. Not sure how many people want to use a chatbot to find basic answers about our Church on SMS, FB, Slack, Twitter, etc. Since this info can be easily found on our website or App.
Giving seems like it could be a good application for this.
In the Rock community we’ve worked on a couple of items.
We have a couple of Slack integrations with our community. They do things like point tallies (@narido ++) as well as allow us to automate / schedule the communication into Slack. We also keep a log of the conversations going on in Slack with a custom bot.
We also have the ability to launch Rock workflows from SMS which allows churches to automate processes like signing up for classes and engaging in content. For instance we used it recently to have people text in to have us look up which small group region they were in based solely on their phone. Others have used it to develop full conversation based chat bots.
Couple of thoughts:
The Slack integrations can be made very powerful, but these tend to be more for staff as it’s difficult to use them with attendees. We tend to be more interested in church → attendee interactions than church → staff.
Chatbots with attendees are great for event sign-ups, setting up giving, registering for content and other administrative needs. I worry though that anything beyond administrative tasks would feel a bit ‘impersonal’ when it comes to spiritual needs. Airlines and banks can probably get away with that more than a church. All that said still a lot of ideas available to implement.
Agree on both fronts. Seems like a great way automate event sign-up, finding info, some baseline support, giving, etc.
One of the things I’m trying to process through is the mediums. I agree that slack seems like a very poor ROI tool for attendee interactions.
FB messenger is obviously an option for many people due the wide spread nature of FB accounts. But I don’t know if pushing people to FB messenger is a strategic move I like. Something about it seems inconsistent and out of place. Just my initial gut feeling.
Same for twitter but even more so. Less people with Twitter accounts.
Email seems like a poor option as well because it’s not quick and doesn’t create the environment of a back and forth convo.
Web-based seems like it could be really powerful but a lot more work than a lot of the other options.
Building it into an App seems like the best use case but also one of the most resource heavy options.
SMS I’m on the fence about. Would love to talk to more people who have done this. We use texting for Smart giving which works well. This currently seems like best entry medium into doing a chatbot.
Yea so the chat bot portion is for basic informational stuff long term goal is to off load calls and bulletins onto the chat interface, our app and website. And yea they will text the number to initiate a conversation.
So api.ai is able to determine preset intents of the user. So if they text something like “Can you pray for me” we can respond by asking for they’re prayer request. Their response is then sent to our prayer endpoint. Long term we also want to be able to match them against our ChMS.
Text giving is two fold. So if their number isn’t set up for text to give on our accounts page, it will shoot them back a link for onboarding. If they are set up then we just take the money using Stripe.
I really like api.ai. It’s a great for basic NLP without a need to feed it a large data model. For large data processing and ML it’s not the greatest because it’s more of a black box. But it’s great for our current use cases.