Comment by kowdermeister | Hacker News Reader

Comment by kowdermeister | Hacker News Reader

Parent Full thread

Maybe I'm not the exact target market, but I always carefully pick my search parameters and research my decision making. For showing up, this is impressive to send flowers, but there's always some minor details that break the experience, like the shop is closed, I can't pay via X service. The Hotels.com booking is great, but not everybody is that rich to book a villa :)

As I see it, Viv aims to be the Google of service providers which is a pretty neat goal, nobody did that successfully, so I'm rooting for them. The platform approach is also a good marketing strategy that they want to allow it to embedable. For example letting people say ask basic things about a monument in a remote national park. The applications are really limitless if it fits on a Raspberry PI like card.

As someone who like to break things, please Viv process this for me:

"VIV, please find me a cheap flight to Vienna, check Ryanair and Wizzair on the 21st of may for two adults" "order the results by price" "show details for the second one" "show the previous one" "are luggage included in the price?" "book it, but skip all marketing offers from the company" "find AirBnB apartments for that day that has wireless internet" "order results by price" "ok, then search in the range of 10 to 50€ per night" "select the first place" "message host: looking forward to meet you" "book it." "Viv, bring me a beer"

I feel the same way. I actually like dealing with the details and making sure I get the absolute best experience.

Wifi? Best price/date compromise? Probability of on-time departures? Opportunities on connecting flights? In-flight alcohol? In-flight chargers? Free meals? Expected leg room? Luggage price? Carry-on limits? That's a 5 minute conversation right there... 20+ minutes if I want to explore results. I'd call a human if I wanted a conversation.

Voice UX is about commands. Fulfill my command if you can give me good results. If you can't, send me to a website pre-filled with data from the command and let me do the rest myself. Anything in between is just a reinvented phone tree.

> If you can't, send me to a website pre-filled with data from the command and let me do the rest myself. Anything in between is just a reinvented phone tree.

That's a really neat idea. Unfortunately it seems to be completely unscalable. For instance, booking on an airline site usually involves navigating and fill a complex multi-page form. Some of the fields are privacy conscious information. The form layout and process varies from airline to airline. Those are just two of the biggest concerns.

How is it unscalable?

"Book me a flight to London on Monday" goes through a speech to text recognition service, then NLP translates that text into service queries. At some point in that chain you're querying a flight aggregation service (ie Kayak).

Just slap together an SPA that calls the same aggregation service, and prefill it with the recognized query.

Amazon's Alexa app is already pretty close, although their queries are less sophisticated (3p integrations just get the literal speech 2 text translation, no deep query transformations/context at play). Google Now is even closer... ask it to "Set an alarm tomorrow morning" and it will give you a partially filed out form with the option to finish it vocally or visually.

There are several reasons.

1. The general pattern of an aggregation service only covering the lowest common denominator features among what it aggregates.

2. I'm not sure of Kayak's terms specifically but accessing via third party aggregator API vs. direct API can affect who gets credit for the referral (i.e., $$$).

3. At the aggregator level, _someone_ has to write the glue code for all of the services. See IFTTT for an example of this. This is a good foreshadow of the problems here by analogy to IFTTT [1].

4. "Just slap together an SPA that calls the same aggregation service, and prefill it with the recognized query." Who's going to build the SPA? For each service vertical? For each company?

[1]: https://blog.pinboard.in/2016/03/my_heroic_and_lazy_stand_ag...

1. That is just convention, services are free to filter on features that are present on a small subset of the data.

2. I don't know how Kayak specifically works, they're just a general example of a customer-facing website that also vends an API that could be integrated with Voice UX.

3 / 4. Everyone has a different approach here and I'm not sure how Viv's would work... but Alexa's ownership model is app-based. 3rd parties build services that process the user's command however they want. They build the backend service, the Alexa voice integration endpoint, and (optionally) a companion website. Amazon owns the marketplace and physical interface. In this model, Uber owns the Uber integration, Lyft owns the Lyft integration, and whoever manages to search both Uber and Lyft would own that meta-service integration.

Viv is being presented as a platform for other developers to build on top of, and these 3rd party developers will be the ones responsible for fleshing out the specifics of how the queries will work given the domain they are implementing for. They will be the ones who will need to anticipate all of the different ways a user can request for something in particular and include all of the edge-cases that might trip-up the API.

In the end I think queries for certain types of well-established services like asking about the weather are probably going to yield better results than more obscure/complex things like placing an order for building custom furniture from a small workshop.

I'm aware of that, but will people realize that? They will just say Viv sucks despite the Netflix app was sloppy.

It's also not clear that what went wrong in certain cases. In one request part of the parameters can be handled by Viv core, part by Netflix and part by another app.

Yes, but who is in control of whether "find me a hotel" searches hotels.com, or Kayak, or Expedia, ...

I think 80% of commands will be handled by Viv's core app, then the third party apps take control seamlessly. It will also learn that you always prefer AirBnb over Hotels.com.

Whatever the default search target for travel stay is, if there is one, will probably determine a large percent of what users will simply stick to using without bothering to change it to another preference, since they may not know better. This default does leave an opportunity for some companies to have an edge over others, and potentially pay to make it so.

Not if there's only one intergration.

The problem with third party integration is that you're going to end up getting a bunch of duplicate "skills", to use Alexa terminology, of varying quality and capability. It's the App Store problem, and bot store problem all over again.

Reply on news.ycombinator.com