M — the virtual assistant
So, I’ve decided to start blogging about products. Why? Well, the story starts with how I got to Silicon Valley. I started working at a VC as an analyst in bay area after college. No, it wasn’t as glamorous as it sounds. The VC was small and still in early stage, but working there gave me an opportunity to interact with entrepreneurs, and I was always excited to see new products come out everyday. After that experience, I realize I want to get into Product Management, and learn more about how to build a great product. This week, I started my master in software management at CMU. So I figure, why not start writing about products?
So, I decide to dedicate my first blogpost to M by Facebook Messenger. Last week, Facebook messenger launched M, a virtual assistant that does (wants to do) everything for you from calling Comcast to ordering takeouts. And luckily, I got it on my messenger the next day (Yes, all of a sudden I feel like I am one of the cool kids in school again). With all the curiosity and excitement, I started using M right away.
So the test drive starts
As soon as I agreed with the “M terms”, M popped up on the sidebar, and sent me a message right away.
Sometimes M will ask follow up questions to refine the search results.
Response time
So how fast does M respond? Once I asked M for restaurants recommendations, and it got back to me with 3 options within 5 minutes. Not bad.
Overall, based on my interactions with M so far, it usually responds within 10 minutes on average, or faster. Personally I am fine with the speed because it just launched and I don’t really have a high expectation at the moment. Eventually people will expect it to get much more faster, but I think after the algorithm collects more data and starts working, M will be able to respond more quickly.
Lack of integration
When I asked M to make a dinner reservation for me via OpenTable, it didn’t work because it couldn’t use my account on my behalf. Well, I was a little bit disappointed. But I can imagine in the future, we will be able to just login directly via Messenger (like how we usually log in with Facebook in other apps), and then either Messenger will re-direct us to the app with all the information filled in for the reservation, or M will make a reservation for us after connecting OpenTable to Messenger.
But this won’t be just for OpenTable, a lot of other apps can be integrated into Messenger so M can do more.
Not enough options
The other thing I think could be better is that M should suggest more options every time. For example, when I asked M to find me some websites that I can buy birthday presents, M only gave me one option. This is not the first time, and often I will have to ask M again for more options. And this could go back and forth for a while before I find something I want.
I feel like for users like me, I would want to have 3 -5 options to choose from. Of course I do understand some people might prefer to have more options to compare while some prefer to only have one so they don’t have to think about it. So I guess the best way to do this is M should learn about our behaviors during the conversation, not just what kind of things we usually ask about. With the algorithm, M should be able to identify if a user only wants one recommendation or more. In that way, it can save both parties a lot of time.
So…what is M really?
At the moment, M is still training its algorithm to get better and there’s a team of people from Messenger working on this to meet all the requests, but we can foresee that M would become a very powerful tool if Facebook got it right.
Overall
I am very happy with my experience, and I am still using M on almost a day to day basis. In fact, I am moving into a new place soon, so I can’t wait to use M to set up Comcast for me ;).
Thanks for reading. Feel free to leave comments and tell me your thoughts :)
PS: I also stumbled upon another article about M on TNW. If you would like to learn more about what M can do for you, you should definitely check it out.
