Five Answers About The Future Of Chatbots


By Beerud Sheth

Founder & CEO | gupshup.io

April 19, 2016

At Gupshup, a bot building platform, we’ve been working on bots long before they became trendy, so I’ll take a stab at answering these questions.

  1. Not much and they don’t have to. Platforms won’t offer much more than APIs, and I think, they won’t have to. There is already an ecosystem of players emerging around the messaging platforms. Facebook is unique among messaging platforms in that they’ve made heavy investments in AI. Other messaging platforms are unlikely to. More importantly, there are many independent platforms (api.ai, luis.ai, even Facebook’s wit.ai) that can be used with any messaging platform. Even before we get to AI, there’s a greater need for tools to support the end-to-end bot lifecycle (dev, test, deploy, host, publish, monitor, track), which companies like mine (gupshup.io) are addressing.
  2. Probably not. Messaging platforms have more ways to monetize than apps did. Unlike apps, messaging platforms can push content. Businesses will pay for the ability to push content (with optins). Messaging platforms are already experimenting with sponsored messages (contextually). The 30% tax didn’t apply to e-commerce transactions on apps, and won’t apply to bot-commerce either. So, maybe the x% of payment may apply to micro-transactions and bot subscriptions. My guess is that the monetization models will be closer to websites than apps.
  3. While websites and apps made humans behave like computers, bots will make computers behave like humans. In that way, interfaces will certainly become more natural. Bots will be more forgiving and understanding. But the dependence of bots on NLP and AI, while certainly a good thing, has been way overstated. For example, structured interfaces aren’t going away. In many cases, it’s easier to point than type. My company built the Teamchat app with structured messages, and Messenger, too, introduced cards this week. In other instances, a precise word is more efficient than natural language — there are many examples of bot builders actually rolling back AI/NLP capabilities and going with the “less is more” approach.
  4. A better question is what will bots not do? To me, bots are a once-in-a-decade paradigm shift. At the very least, it’s a new front-end. So, everything we do on a website or an app, will be repurposed into a bot — too many use-cases to enumerate! Since mobile devices are already ubiquitous and bots will be more ubiquitous than apps, bots will impact more aspects of our daily life than websites or apps ever did. We will also have new categories of bots that are only possible in the messaging paradigm. (That’s a longer discussion). Even without AI, the bot ecosystem can be as big as the web ecosystem. AI and NLP opens up more possibilities in certain contexts.
  5. Yes, paid bot distribution is inevitable. Search and discovery of bots will be a challenge soon as you have millions of bots. Bot search engines, bot stores and reviews will come. Sponsored messages and paid referrals will inevitably follow. However, I don’t expect bot distribution costs to get as high as they did for app installs. The real problem with app installs is that users just don’t. Most users used no more than a dozen apps every day. Consequently, the CPI rates were high. It’s far easier to consume lots of bots (or websites) than it is to consume apps. I expect bot distribution economics to be closer to the web ecosystem than the app one.

Yes, it is indeed an exciting time. Bots are the new apps. Or, maybe, the new websites. Ok, ok, metaphor overdose! Let’s just go build bots and see what they really are.

Leave a Reply