I wanted to share a story about why I'm so excited about messaging.
It's 2009, and I've been at Best Buy 7 years since selling Geek Squad to them. I'm sitting in a meeting where some PR firm is pitching their next year's projects for Best Buy. The bid? 3 million. As I was oft to do, I'm on twitter - watching pissed off customers tweet one problem after another. I turned to Best Buy's CMO at the time, Barry Judge and said, "What if our PR strategy was to listen to customers?" That's when we took a small piece of that money and built an in-house twitter tool to allow customers to message customer support. It was similar to a custom version of ZenDesk. The vision was to imagine EVERY employee of a company accessible through real-time chat, use the data to predict problems before they occur, and empower employees to help. The most people we were able to get on that system was 5,000 out of 150,000 employees. I left Best Buy in 2012, and sadly there are likely less than 5,000 still using it.
We are excited to try and accelerate 2 things: expose more APIs that allow for customer self-service, like scheduling, pre-ordering, queuing, requests, etc. Further, when you need to talk to a business, that you can do it through a single interface where YOU have a copy of the conversation "for quality purposes".
We all have a long way to go, but I hope you can see what kinds of service could be delivered in the coming months and years.
Very impressed with the smooth Facebook Messenger connection. It just works, no installation required.
I've never seen a bot working inside Facebook Messenger - is this an official connection, or are they going to shut you down when the word is out? Very interesting!
Can you share more info on this?
@mmariansky can't really say much but I think it's obvious where all this is going. I started Geek Squad as soon as I saw the web in college. I haven't been as excited until now in hopes of using messaging versus voice for customer service.
This is so cool. Not sure if you guys can answer this yet publicly, but I'll throw the question out: what type of AI are you using? Did you build your own using deep learning algos , or is this a machine learning/Watson type set-up, or are you just licensing some other AI? How much of the guts can you tell us about?
@tuckermax As bots become more and more common, machine learning and technologies like recurrent neural nets will be the key to expanding capability through messaging and text interfaces. There is a lot of very new open source code being released by Facebook, Google, Microsoft, Apple, Amazon and others. This is one of the keys how we plan to improve @Assist.
I also think it is no accident that Mark Zuckerberg chose to build his own "Jarvis" for home automation - which will use text and voice control. Think of the message this sends to all Facebook engineers, especially the Messenger and WhatsApp teams.
@tuckermax Thanks brother and thanks for being in my camp for the past year. Much appreciated for all your thoughts and support. I don't care what people say, you ain't too bad of a guy.
I'm really getting tired of emails for notifications. I much rather have text messages for everything important... I will definitely be using the crap out of this. 🙌😀
@sachagreif I think you can expect notifications and preferences to become much smarter and learn what you like and don't like. That's what's different this time.
So TechCrunch just published: Assist Has No App. It’s A Free Travel/Shopping Chat Bot For SMS And Fb Messenger h/t @joshconstine
Assist is a chatbot that aggregates APIs to let you message GrubHub, Eat24, Postmates, Uber, Lyft, OpenTable, StubHub, and more.
It works across Facebook Messenger, Slack, Telegram, and SMS
I personally think 2016 will see an influx of Chatbots and what a fantastic one to kick off the year with!
Plans include going into suggesting services for users which would be really cool!
I'll let @shanemac and team tell you more :)
Where does it currently work/work the best and where are you looking to expand next?
What is the story here?
@bentossell I will tell you that v1 has one simple goal: take all of the local services that you could order with your mobile phone and access them without installing an app if possible. Next up is to try and find and integrate as many services that we all know will be emerging in the coming years.
I find it particularly interesting that this is available across a variety of popular platforms — SMS, Slack, Facebook Messenger, etc... Currently, other products in this space are tied to one (Magic via SMS, M via Messenger, Operator via its own app).
@shanemac — What's the reasoning behind this? Do you see people using Assist differently on each of these different platforms?
@jackdweck We expect this will be the case and we are keeping a close eye on it. Specifically the difference between enterprise and consumer messaging platforms. The other thing is we anticipate the platforms to open up messaging UI and languages at different rates so consumers might prefer better experiences if some of the platforms allow you to better purchase, view content, etc without using links or command line level type interface. Many people prefer design and it is coming.
@jackdweck One of the key reasons (I suspect) for Assist not having their own app, is that app fatigue is a real problem, and the reasons that people are now flocking to building on top of messaging is because it lets them go to where their users (or potential users) are instead of getting the users to have to come to them. Accordingly it makes sense to support multiple platforms. Fast forward to next year and I think multi platform support will be table stakes.
(Not taking anything away from Assist, it looks very cool, and I think Magic et al should be watching their back)
@tonylucas all true. An additional reason for no app is we really tried to use the harsh constraint of SMS to force us to see what we could do without the luxury of device or OS APIs. A concern is: will there be too many messaging platforms to make bots cross-compatible?
@shanemac@jackdweck Another great example of the shifting landscape in product interfaces; users want more conversational products and until apps / software becomes adaptive, messaging platforms such as SMS, M and Slack are going to become more popular as the base of new apps. Do you guys have any concerns that people might stop using SMS or messaging in the near-term, in the same way people abandon other apps?
@_pulkitagrawal@jackdweck SMS is still ultimately controlled by the carriers, so nothing would make us happier to have a better, more capable universal platform. Are we worried that FB or Apple will run this show themselves? If I've learned anything in tech, it's that there's always room for a ton of players when the field is this wide open.
I'm actually more worried that developers/companies will have to code their bot for each large messaging platform. This works against the dream in the first place: less work than an app.
I think you have some troubles on message deliveries. There is some delay (about 1-2 minutes) for each answer.
At this time, it's faster to download a specific app and order with it than using Assist.... which is not your honorable purpose. Keep up the good work.
@charloganem oh my god, there is soooo much to be fixed, improved, debugged, etc. We wanted to get it out there as soon as we could on Product Hunt to show the community partly in hopes of getting feedback as well as interest from engineers who want to work in messaging and machine learning.
Just tried to book a haircut in midtown (23rd /3rd area) and was given a location in Bayonne, NJ (32 minutes away in a car). Not sure it's quite ready for prime time
@rstephens@walterareid Well, based on that response, as proof of concepts go, this is pretty good. Not sure what will differentiate this from the 10 clones that are waiting in the door though.
@walterareid if there are only 10 clones, then I'll be very disappointed with the Product Hunt community. When I started The Geek Squad there were (and still are!) thousands upon thousands of "competitors". Rising tides...
@rstephens@shanemac congrats on getting this out there - do you anticipate being able to predict pricing on services in an area for services with enough transactions, e.g. "If you need a cheaper but good X, try y service that is similar to the more expensive version you searched?" or will selection be opaque to the user for now?
@grmeyer I think not only could you predict pricing, but more importantly, you could help retain preferences across services. Right now, only Delta knows I prefer a window seat. Only Westin knows my hotel preferences. It gets super-interesting when you aggregate category preferences across user populations by geography, weather, etc.
@rstephens Excited to download. P.S. one of my favorite quotes is still this one from you in 2008-ish: "Marketing is a tax we pay for being unremarkable."
I guess I can't use this yet since I'm not in the US (any plans to roll this out in Europe?). But the approach sounds very promising. Just out of curiosity: what are your thoughts on Magic, GoButler, Facebooks's M, Siri (?), etc., in particular in comparison with assist? Also, what's your long-term vision with this?
@wholeearthweb At this early stage, the more services, the better. Magic and M are great for complex, human-powered tasks. We chose to focus solely on non-human API services. Not every business can afford to build an app, and we don't believe people want to download an entire app just to book a haircut appointment. (Assuming they can even know about these apps in the first place". That's where messaging comes in. Now imagine where a non-technical shopkeeper can literally "program" via chat to expose their inventory, appointment scheduling etc. Assist wants to be the easiest place to access all of those new services.
Just installed it. In case you're interested in bugs (I know you realize it's still buggy): I downloaded it on my iPhone for FB messenger, then got redirected to the AppStore, then opened messenger from there and nothing was there. Then went back to browser to your site where I saw a request to open messenger. Then I saw assist in messenger but no instructions on what I do next (played around with it then and got a response, but just as a feedback for user onboarding).
@wholeearthweb as a v1 - it's likely buggy all over the place. we DO have a feedback method built in: WHile in any chat on @assist, on any platform, just begin your feedback with !
That will not interrupt your current session or task. For example: ! FBMessenger has a bad redirect to the App Store...
@wholeearthweb you can still use it. I'm in Canada and when it asked my location, it worked fine. Although the recommendations probably aren't as refined, you can still play with it :)
Congrats Assist team! Super excited to see this out in the world.
I'd love to hear you share a bit about how you ended up with this product. Assist has gone through a few different iterations and pivots over its lifetime. How did your journey land you here?
@davidspinks Great question. When I saw the pre-assist screen shots, it was a local peer-to-peer travel chat/advice product. There was one feature that stuck out where as a recommendation card was sent to another, it appended services like Uber & Hotel Tonight right in the card. I encouraged the team to first focus on that feature and flesh it out. We all know that there will be more and more APIs for services - and that's where @Assist is today. It happens to be part of a larger trend where basic UI like SMS & text chat can be great enhanced by APIs.
Add in machine learning over the next few years, and I think we all will have even more capabilities. How long does it stay in SMS or on chat platforms? My guess is all of the above, including notification layer, Spotlight Search ios, and streaming apps from Google.
Feels like 1994 all over again.
Hi there, I am part of the Assist Engineering team, just wanted to let you know that we have significantly improved the Facebook Client implementation, it should now be as fast as the rest of the channels. Please try it out and feel free to send any issue on my way!
@davidiwanow I think an interesting idea is what Apple is doing with spotlight search and Google's new concept of "streaming apps" as well. Similar to how many others have said that the notifications layer can have "app-like" properties.
great product - just a tad slow but great potential. Any challenges you came across the way? Preferred technologies/platforms used? Any quick wins you can advise a noob in the space - i'm just getting my feet wet in chatbots and trying to soak up as much as I can. Again...great job!
Do you ever see it making sense for me to have an Assist account where I can store my credit card details and preferences for services like my home address? e.g. "Can you call me an 'Uber' to take me 'home'"?
@nhallam for sure, that is kind of the magic of it all. Preferences across all these services. We already do this with addresses, personal information, not yet payment. That said, Stripe does save your card across services so that works perfect... but you are spot on, it's coming.
@shanemac Radical. A friend of mine is a heavy user of Pana and has a lot of good things to say about the service. I liked how there is one main conversation, but they chunk them historically by topic or enquiry. Feels like you need your own application to make that kind of thing possible. I can see a use case for when I want to go and see a list of all my existing enquiries or receipts etc. but I guess all that can come down the track. The horizontal play feels more natural, but maybe more difficult to execute well on. Hope it goes well.
So awesome to see the shift from the original app to an interface on messaging platforms. Makes SOOOOOO much more sense! Will have to try it out again with this new interface :)
Replies
BotWatch
Shisharka Land
Doodle
BotWatch
Muddy
BotWatch
BotWatch
Going Deep 123 - Tucker Max, Book-in-a-Box founder
BotWatch
Simplecast
Sidebar
BotWatch
Makerpad
BotWatch
Makerpad
BotWatch
Makerpad
Unroll.Me
Simplecast
Converse
BotWatch
Helpbar.ai
BotWatch
Drop
BotWatch
BlueLight on Apple Watch
BotWatch
BotWatch
BotWatch
Simplecast
BotWatch
Sketch Hunt
BotWatch
BotWatch
Deliriously Happy
BotWatch
finimize
BotWatch
finimize
BotWatch
BotWatch
CMX
BotWatch
Assist
BlueLight on Apple Watch
ContentKing
BotWatch
BotWatch
BotWatch
Broadcast
Simplecast
BotWatch
Simplecast
The Nifty