Natural language is the general purpose “API” for humans and it seems it may become the same for software. Bots are all the rage on several platforms including Telegram and now Slack. Facebook is experimenting with M. There are bots to schedule meetings. Bots to order lunch. And so much more. Our analyst Jonathan has wrote a great post about this development.

Many of these bots are being built by new startups that are raising venture capital. This is fantastic news with regard to advancing human computer interactions and may also lay the foundation for breakthroughs in AI (more about that in tomorrow’s blog post). So from a social perspective it is great to see capital flowing to these bot companies.

But the competitive dynamics and who will create sustainable businesses are a lot less clear. Will it simply benefit the underlying platforms which could attempt to suck the best bots into their platforms? Or are there enough platforms that we will see cross platform ubiquitous bots and people will want to interact with the same bot across all of them?

Where could a sustainable competitive advantage for a bot come from? Most bots are building some kind of machine learning system and so having done more learning than others is one possible answer. The challenge with that answer though is that all interactions that train the bot are also visible to the underlying platform (and possibly even to other bots on the same platform). So that’s not super compelling.

A way around that is to launch your own app that has a text based interface. Operator is an example of taking that approach. If your own app succeeds then you may have access to training data that others don’t. If you grow your app big enough and get good enough, then you can eventually make that same bot available across platforms where the platform and other bots may be listening also.

Another place to look for sustainable businesses in the Bot Rush is to consider the picks and shovel suppliers. Is there something that all bot companies can and will use and won’t consider core? One thing might images. If your bot needs to understand what’s in an image you may not want to build that yourself but rather use an API such as Clarifai.

Having written a bad version of ELIZA as one of my earliest programs, I have been excited about the prospect of conversing with machines for a long time. So I will be watching this development closely and appreciate all pointers towards innovative startups (with the caveat above in mind).

Natural language is the general purpose “API” for humans and it seems it may become the same for software. Bots are all the rage on several platforms including Telegram and now Slack. Facebook is experimenting with M. There are bots to schedule meetings. Bots to order lunch. And so much more. Our analyst Jonathan has wrote a great post about this development.

Many of these bots are being built by new startups that are raising venture capital. This is fantastic news with regard to advancing human computer interactions and may also lay the foundation for breakthroughs in AI (more about that in tomorrow’s blog post). So from a social perspective it is great to see capital flowing to these bot companies.

But the competitive dynamics and who will create sustainable businesses are a lot less clear. Will it simply benefit the underlying platforms which could attempt to suck the best bots into their platforms? Or are there enough platforms that we will see cross platform ubiquitous bots and people will want to interact with the same bot across all of them?

Where could a sustainable competitive advantage for a bot come from? Most bots are building some kind of machine learning system and so having done more learning than others is one possible answer. The challenge with that answer though is that all interactions that train the bot are also visible to the underlying platform (and possibly even to other bots on the same platform). So that’s not super compelling.

A way around that is to launch your own app that has a text based interface. Operator is an example of taking that approach. If your own app succeeds then you may have access to training data that others don’t. If you grow your app big enough and get good enough, then you can eventually make that same bot available across platforms where the platform and other bots may be listening also.

Another place to look for sustainable businesses in the Bot Rush is to consider the picks and shovel suppliers. Is there something that all bot companies can and will use and won’t consider core? One thing might images. If your bot needs to understand what’s in an image you may not want to build that yourself but rather use an API such as Clarifai.

Having written a bad version of ELIZA as one of my earliest programs, I have been excited about the prospect of conversing with machines for a long time. So I will be watching this development closely and appreciate all pointers towards innovative startups (with the caveat above in mind).