On chat bots and enterprise concerns

“Bots are new apps”.

It is a bold statement but it’s supported by Microsoft, Facebook, Slack and Telegram.
So you can almost say it’s a new reality. Which means new opportunities for all of us. Let’s dissect the issue of chat bots so you come prepared as the land grab is starting very soon.

Is it novel?
Chat bots are not novel. It doesn’t matter though. Now all the pieces are coming together in a way that’s changing the apps landscape.

Chat bot is AI and needs expensive engineers with PhDs and years of research
Not true.
1) Bots don’t all need to have AI elements – recognising keywords is enough and can be coded in few hours
2) The ecosystem is young but a number of bot building platforms are available already: Chatprime, Pandorabot and Chat fuel are just some of those, Slack already has own botkit, Microsoft is releasing its framework and Facebook is expected to do so in near term.
3) Once you have your logic done it’s actually easier to implement an average chat bot compared to a native mobile app as you leverage the existing interface of a messenger app.

Why now?
It seems all the foundation building blocks just got hardened enough for the ecosystem to emerge.
That unleashes the advantages messaging always had:

– Natural: humans always liked interaction that is: 1) instant; 2) natural.
– Accessible: chat apps have billions of users, you’re likely using one of them on daily basis
– Light-weight: having to download another app with custom UI for every pizza shop is not what users want, apparently. Engaging a bot is as simple as messaging someone or adding a contact into a group chat.

I built a bot – what next?
You still need a messaging app to run it on. You’ll need to add your bot to Slack directory, Telegram store, Facebook store, Microsoft directory etc.

Messenger protocols usually are similar but implementations vary so you may expect having to publish a different bot for each messenger app. This should be less problematic than building same app natively for iOS and Android though so once you’ve coded the logic it should be fairly easy to adapt the connectivity to multiple platforms.

Similarly to app stores, bot stores are going to make your product discoverable but you still have to do your home work in regards to marketing.

Future of the bots, what is it like?
I’m thinking about writing a separate post on what future looks like for bots and messaging in general. In brief, these seem to be the major evolution vectors:

– Better AI, more autonomy, more complex tasks, better NLP (natural language processing)
– VR and AR (Virtual Reality and Augmented Reality)
– Voice recognition and Brain interfaces
– Video chat bots
– Cross-platform penetration
– IoT connectivity

What about security and data ownership?
The concerns that enterprises have in relation to messaging aren’t going away with chat bots.

Many businesses aren’t prepared to host their customer data with Facebook or Microsoft. Enterprise CIOs tend to favour on-premise dedicated server setups over co-hosted cloud solutions and they have good excuses for doing so. Custom trust & safety protocols, data protection policies, industry-specific standards in health care and finance; customisation, white-labeling, custom SLAs, you name it. These won’t be addressed by Skype or Telegram which pretty much blocks bots from revolutionising the enterprise segment.

To solve this problem as an enterprise you could build your own messaging app – from scratch, using one of the open-source XMPP servers such as ejabberd, Tigase or Openfire for your messaging backend. Alternatively, you can license it from one of the CaaS (“communication as a service”) providers.

The options you need to look for:

– Dedicated server – you need to have full ownership of your data and full control of your infrastructure so opt in for private cloud / on-premises set up with full access to application and databases for your tech team
– Close-source messenger client – GPL doesn’t really work for many commercial projects so make sure you choose a platform that is able to license under Apache, MIT or a custom commercial license that doesn’t require you to open-source any modifications you make

At QuickBlox we provide both shared cloud and dedicated set-up options from day one simply because enterprises need this kind of security and are willing to pay more for it. This allows us to re-invest into platform R&D and consequently improve the product for both free/small-cheque users (developers, startups) and enterprise, so everybody wins as a result.

The concept of allowing dedicated / on-prem install is usually frown upon in the world of SaaS and cloud. We managed however to find a model that retains the benefits of SaaS/cloud with ones of dedicated server set up. There is no contradiction – with the help of technologies such as AWS Cloud Formation and Docker it is perfectly possible to deploy, serve and upgrade thousands of dedicated servers at once if needed while continuing to leverage the SaaS model.

I like the “Messaging OS” term. Not sure who has coined it first but I first stumbled upon it the Mary Meeker’s “Internet Trends 2014” report (and this, too) and then saw couple mentions in TechCrunch as well and I loved it as it explains best what we’re working on and the added value we create for the market.

Now it is all clear. Market has evolved a level of abstraction higher.

Messaging = new OS

Bots = new apps

It’s like with programming languages when you move up levels of abstraction from binary machine code to Assembler to C to Object-oriented it becomes easier and more fun to build useful things. Doing less work you create more value. This frees you up to explore the unknown and innovate.

So, have fun innovating with bots. Just ensure they obey those Asimov laws, will you 😉

P.S. check out a video of a simple chat bot I created back in 2014 for switching home power socket on and off (in addition to Q-municate open-source messenger app it also uses Raspberry Pi and Powergenie power sockets):

Github source code:
https://github.com/QuickBlox/sample-powerbot-python-rpi and there is a Rust sample too: https://github.com/QuickBlox/sample-chatbot-rust