There’s a bot for that? — Thoughts on bot discovery
Seems like the current hype around bots and conversational UI is widely seen as the biggest thing since the App Store launch in 2007, and I can easily relate. We’re already addicted to communicating with everyONE we know via messaging apps, why not copy this user experience to communicating with everyTHING we know? Smart bots can get stuff done for us (schedule meetings, turn on the AC before we get home) as well as help us deal with brands and commerce (change data plan with your carrier, order toothpaste on amazon) without having to leave our favorite messaging app, and do so more efficiently than before. These new technologies and experiences will no doubt bring many new paradigms, which I’m very curious to see pan out. The first one I’d like to discuss here is discovery and distribution.
Searching Google for “bot discovery” currently brings back 1630 results and apart from an interesting piece by Vijay Sundaram about Slack bots UX, no one has really covered this topic yet. We’re still so early with bots, looks like everyone is still figuring out how and what to build, and has yet to face the discovery challenges.
With bots, unlike apps, there’s no downloading or installing required, and starting to use a bot require zero resources from you — the bot is in the cloud so it doesn’t take any space on your device, it’s just another open chat dialogue which takes milliseconds to start. Since there’s so little friction with starting to use a bot and bots vastly reside on messaging platforms it makes sense that discovery will be hyper-viral. Here are a few discovery flows I can think of:
- Friends recommend a useful bot they used — “you should check out X.ai, they have a great assistant bot for scheduling meetings” and a link to the X.ai bot is added automatically so you can click to start a conversation with it, or better yet — the X.ai bot starts a new conversation with you to introduce itself and you have that chat available if you decide to give it a try at some point. This is the fastest possible form of word-of-mouth — getting a recommendation about a service or brand automatically launches a conversation with that service or brand.
- Friends invite a bot to join your conversation — imagine you’re talking to your sister about buying your mother a birthday present and your sister invites @amazon_bot to suggest some ideas, you pick something together and check out. From then on you “know” @amazon_bot and its there at your disposal.
- Bot Directory — something similar to the contact listing when starting a new chat only with available bots we can choose to talk to.
- Bot Store — something that looks like an app store for bots. You can already see examples of this on Slack and Kik.
- Bots introducing other bots— just like the two first flows, a bot can recommend other bots or invite other bots to join the conversation. You can see this happening on Luka as well as Telegram’s Store Bot (a bot for discovering other bots)
- Bots mediating other bots — when you think about it, do we really need a bot to introduce another bot when we could theoretically have the bot we’re currently talking to simply deal with that other bot for us? For example: you ask an assistant bot to get you a large pepperoni pizza, and that bot doesn’t have that capability but knows (or can look up) another bot that does, or maybe even multiple ones that do. In this case, your assistant bot can take care of communicating with those other bots to find you the best options, ask you for your preference and then go back and seal the deal.
This last flow is the one I find most groundbreaking. If bots mediate other bots for us, we could find ourselves doing business with brands seamlessly (for example: “I got the game tickets half price, not sure where my bot got them though… I’ll have my bot talk to your bot”). In terms of discovery — if you build a bot that’s useful to other bots, it could “go viral” among bots and could have massive reach seconds after launching.
