Last weekend in Pune, India, Fresco Capital along with our longtime partner, e-Zest, produced a 24 hour hackathon about building bots. A few weeks ago, I explained why a Venture Capitalist is running a developer hackathon in India. Our main goal was to learn about bots by seeing what developers are currently doing with bots and using that to look into the future of bots.
The Current State of Bots
At the hackathon, we had over 125 developers pounding away at the Facebook, Microsoft, Slack, and other bot frameworks. After looking at over 50 bot applications, we learned a lot.
The most common thing that stood out was that most of the users interacted with the bots in some form of chat client. Facebook was by far the most popular interface for the bots, however, we saw a lot of Slack, Microsoft (Skype), and even a few using Hipchat. One team wrote their own chat interface to interact with the bot.
The second common thread was that for these chat interfaces, most apps integrated some form of natural language processing (NLP) into their interface. Very common, but not nearly as ubiquitous as NPL, was a voice interface.
We also started to detect some common categories of bot applications. While not all of the 50+ bot applications we saw fall perfectly into these three categories, the most common categories are:
By far the most common bot category, we saw integration with many 3rd party tools. For example, Hotel booking with Trident Hotels API in Slack, Skype integration with an internal timesheet application, Pipedrive integration with Slack, Glassdoor integrated with Facebook Messenger. One very creative bot consumed the API of the host and gave you many useful statistics about the conversation thread you are in. These are the most common bot application as 3rd party integration is the logical use of bots based on the current technology and user comfort level with bots today.
Interaction with Hardware
We saw a few bots, including the overall winner, interact with hardware such as the Raspberry Pi. Still interfacing via a char client, but controlling external hardware. This is part of the future of bots, allowing a bot to interact with your TV, music player, and car.
While you can build bots with traditional software development tools, we saw a few tool oriented bot applications. One of the finalists was called "Magic Bot” and they would build your bot for you if you gave the tool your API and a list of commands. We also saw some home grown interaction clients that would also learn you behaviors. Clearly these developers view a future where everyone will be rushing to release a bot, similar to a time where everyone wanted a web site or mobile app.
Startup and Developer Ecosystem
Our interaction with the Indian ecosystem was very fruitful. I got to meet over 25 startups at the Startup Pitch event that was co-located at the Hackathon. Big trends were team collaboration tools (not surprising in a market known for remote development), health care, and consumer based apps. Startups seem to have access to early stage capital, but mid to later stage capital is hard to find. Indian startups can be categorized into something that is either hyper local or something that is very global from the onset. I was not expecting to see such a mature ecosystem and was blow away.
We learned a lot at the hackathon and it was a positive experiment. Look for our next experiment somewhere around the world. :)
Page rendered at Tuesday, 21 February 2017 02:27:59 (Eastern Standard Time, UTC-05:00)
The opinions expressed herein are my own personal opinions and do not represent
my employer's view in anyway.