Democratizing a New Technology - ChatBot Example
I was talking to my friend who runs his own Startup. He mentioned to me that ChatBot was a HOT investment within VC community until FaceBook decided to drop its ChatBot initiative.
Interestingly we saw following two events which happened around the same time
1. FaceBook closed down its virtual assistant ChatBot
2. American Express acquired Mezi (Travel Assistant ChatBot) for a ridiculously high price
These two events have some correlation in terms of technology maturity, and technology adoption.
ChatBot uses IM, NLP, NLG, Machine Learning, and Deep Learning. IM captures the input, NLP translate that into an intent, intent fetches relevant data, and NLG communicates back to user. Machine Learning keeps training the whole thing to reach the relevant intent quickly. ChatBots are definitely promising, and they will be part of our day to day life in some shape or form.
Technologies mentioned above have already passed the hype curve, and getting better every day. However the collective package is not quite there yet.
If we democratize the technology too early before its maturity then the adoption will fail. There will be an element of frustration which will keep people away from the technology after the first use if it does not produce the result.
The traditional wisdom is to solve one niche problem, do it right, and then democratize it to solve other problems. While solving the niche you may need to augment the core technology with some additional technology, and/or human touch. This process helps you to learn all nuances while keeping your customers engaged. You get enough time to learn lessons, course correct, and implement changes. If you lose your customers then you lose opportunities to learn from them.
Mezi focused on one problem. It was created for Travel assistance. It also married core elements legacy IVR with ChatBots to navigate users through options with ChatBot experience. The fundamental platform can be used to solve other problems but they consciously focused on one niche.
FaceBook tried to solve for everything and probably did not see desired results. I am not saying that FaceBook was completely wrong. FaceBook based on it’s business model probably made a right choice to leverage ChatBots for everything. FaceBook was also wise enough to release it to developer community before releasing it for a public use.
If you are planning to use ChatBots within corporate then my advice will be :
1. Test it out for a niche use (e.g. ChatBot for Medical Benefits). Select a niche which is internal before using it for your external customers or suppliers
2. You may want to provide some menu options to get to the intent quickly. You can eventually reduce menu options as ChatBot matures on intent recognition
3. Create a stable process of monitoring ChatBot transactions, and helping those users manually if ChatBot can not resolve it. Engage your business SMEs along with Technology SMEs
Once you nail down end to end process for one niche then expand into other areas.
#AI #ML #Machine Learning #ChatBot #Deep Learning
