Hacker News new | past | comments | ask | show | jobs | submit login
Jarvis: Personal assistant via SMS (jarv.co)
114 points by filleokus on July 27, 2014 | hide | past | favorite | 99 comments



I can't conceive of who would sign up for this given how little information there is on the site. I can't even find the pricing. The FAQ is short, vague, and even feels deceptive (re: "futuristic tech").

Where is the legal? Where are the examples, demonstrations, etc.? No, no, no.


This is a result of the whole lean movement. Validate your idea with customer feedback before you spend a bunch of money. Yes it means things will be a little smoke and mirrors to begin with but that doesn't mean the founders are not capable of delivering what is being proposed.

They may have thought through all of your questions but decided not to have firm answers right now in order to validate the idea, and not the technology.


And therein lies the problem with the whole lean movement. You can never truly invalidate an idea. You can only invalidate your 'lean' implementation of an idea which may or may not have been the actual minimal viable product. It's quite possible the implementation you executed is too minimal. In this case, I think that's exactly the case. The founders have an interesting idea that would appeal to a good chunk of HN readers but we're all put off by just how minimal the implementation is. Unless the founders are reading this thread, they'd easily be duped into believing they invalidated their idea based on their low conversion rate.


This isn't really the lean process though. Lean implies you've done proper research prior to releasing a product. If people are umcomfortable with signing up then this isn't an MVP and not indicative of how the lean process works.


My point is that no matter what you do, you will never be able to invalidate an idea. All you can ever do is invalidate one exact execution of the idea. Perhaps your idea is great but you left out the one killer feature. Or perhaps it's feature perfect but the design is just not right. Or perhaps the implementation is perfect but you didn't market it correctly. Research done prior to launch can certainly be helpful but still doesn't change the fact that you cannot invalidate an idea.

The problem I have with the lean startup movement is how easily it is to get derailed before you ever even begin. For example, the lean startup movement suggests an initial qualitative approach to testing your assumptions, achieved by talking to a small number of people in your target market. But that assumes you know who your target market will be, that you know who the early adopters in that market will be, and that you're asking the right questions. If you get any of those wrong, poof! You'll believe your assumption wrong and your idea invalidated even those it hasn't been.

The founders of Jarvis may have done their homework prior to releasing a product. They may have concluded that there was a real pain point that their service could solve. But what research should they have done to realize that their target market wouldn't be interested unless they had a legal page? Or demonstrations? The problem with trying to go with the absolute MVP is that it's way too easy to go too minimal.

The lean startup methodology will work best for ideas that solve massive pain points. With a massive pain point, it won't matter if you're talking to someone that is normally an early adopter or someone that normally trails behind. It won't matter if you have a poor design, a minimal feature set, and no legal page. But for anything else, it's just entirely too easy for your target market to dismiss you and for you to summarily dismiss your idea.


I've often thought the same. Execution is the hard part, especially with regard to conversions. It is really tough to get right, even with a mature product. And, given the sea of noise that invades everyone's lives, trying to be heard with a product presentation that is purposely subpar seems unrealistic.

So, one problem with lean is that you're actually testing two variables: the idea and the execution. When people don't convert, is it the idea, the copy, the design, something else? It may be that you just need to iterate on your copy/conversion funnel, etc. a dozen times. But, lean de-emphasizes the implementation; encouraging you to instead view the feedback as primarily relevant to your idea, and so move on in the face of low conversions.


All you can ever do is invalidate one exact execution of the idea.

Absolutely agree with you and I wish the folks that teach lean would emphasize this more.

For every successful company, there are probably multiple teams that tried to test the same general idea only to conclude that it won't work.

Overall, I think lean start up movement does a great job of taking credit when it works out while not delving enough into all the people who gave up because the idea didn't validate itself according to the lean process...and yet another company with a similar idea made it work.


> This isn't really the lean process though

That's a bit of a "no true scotsman", isn't it?

Certainly it must be true that encouraging people to belt out a "minimum viable product" results in the launch of something half-baked and unviable at least slightly more often than if the advice hadn't been taken. That doesn't invalidate the lean startup concept, it just illustrates a potential tradeoff... and there's always a tradeoff.


> decided not to have firm answers right now

I'm remembering this for the next question my boss asks me...


They have the same phone # and vaguely similar style to: http://investment1.co/faq


I'd strongly advise you to check out the Cyman System website. The Cyman Digital Butler has a few video demonstrations, and I built it with Jarvis' personality in mind. Might be what you're looking for?


Even the domain's WHOIS has Enom's Whois privacy service instead of an actual company contact listed.


To be fair, I enable Whois Privacy by default on all my domains, no matter what they are for.


Most domains come with free WHOIS privacy for a year, and it costs less that $10 after that.


Tried it. After 3 requests, all of which would have taken less than the prescribed 15 minutes, was told it wasn't the right service for me. Great idea, poor execution on MVP unfortunately.


What was the nature of your requests?


1. Reorder stuff for me on instacart (would have taken 2 minutes), 2. Lookup a URL for me, 3. Lookup 10 quotes. Last one could be called "iffy," but this was after texting them asking for parameters to which they responded "as long as it's legal, we're in beta now so anything goes." (paraphrase) Again, not here to trash these guys / gals, but...


Pretty bold - were those listed amongst that Jarvis' abilities? My assistant (Cyman Digital Butler) has a list of example commands you can try, gives an idea of the abilities Cyman has. So you could say "open the ... website" to get the url for example. Although we've got a great community to help, we're in the middle of creating a wiki. Perhaps the Jarvis you're using needs a wiki too.


"Pretty much anything a personal assistant can do." "Jarvis is a tech enabled person." I would assume from this that all those tasks were possible for a human personal assisstant.


Very true. Far be it for me to defend the competition, they probably are still teaching it certain phrasings. My Cyman assistant is pretty good at understanding a variety of phrases, but I'm still teaching him.


There's no way I'm giving you my personal info without knowing what your pricing structure looks like. It's in the HN title, but it's not on the site.


> It's in the HN title, but it's not on the site

We took it out of the title for that reason.


I also don't see any legal information.


It says $99/month on the site.


"Meet Jarvis, Your Personal Assistant Who Never Sleeps"

Then when I click on Live chat:

"We're not around, but we'd love to chat another time."

Well that's a bit concerning.


Lol, Indeed. I built an digital butler app inspired by Iron Man's JARVIS assistant called CyMan. We've got a great community and dedicated helpers in our Google+ community. Feedback on that community has been excellent! If you ever wanted to give it a try...


Hey - even Jarvis needs a Jarvis!


It's Jarvis's all the way down... D:


I like the idea. Don't know if I would use it but I like it.

That auto carousel is annoying like hell.


Agree that i like the idea of a online PA, but there is too little info on the site, and the examples they give are rather unambitious. I expected to find something like odesk for small but very specific tasks.


Definitely agree with the auto carousel. Not a fan of things disappearing when I'm halfway through reading them.


Oh god, I tried to read the screenshots to have an idea of what is the service like, but I gave up after the second time being interrupted by the carousel. Terrible.


See, I need to hire a personal assistant to deal with email, but I'm afraid of doing the remote PA thing because I don't trust them with my email (and that's a whole other issue since that's the job I need help with) and this idea is appealing, except, as others have noted, I have no idea who this company is. What their business model is. How they make sure my personal details and payment stuff is secure. That this won't just wind up being an easy way to phish into my life.

Nope.


What's the difference between this and Fancy Hands(https://www.fancyhands.com/)?


The two key differences for me:

(1) FancyHands, having been around a lot longer, has far better execution and delivery. Email, phone, an app, recording through the app--many ways to give a task assignment. I usually don't want to text them, I forward them emails to take action upon or simply leave a voice message.

(2) Jarvis is unlimited tasks. THAT is a big one for me. If there ever were a service that were at or somewhat approaching FancyHands quality with unlimited tasks, I'd be on it in a heartbeat. For the time being, I've been using Visa Signature Concierge for tasks that don't require a FancyHands level of attention paid to them, since they are unlimited and free if you have a Visa Signature card.


> For the time being, I've been using Visa Signature Concierge for tasks

Could you explain that a bit more? I have a couple Visa Signature cards, but always thought the Concierge service must be some sort of scam or ripoff. What are some examples of things you've gotten them to do? (When making purchases, do they charge a premium/commission?)


It's definitely not a scam or ripoff, they cost nothing and I use them frequently. I have never actually purchased anything through them, so I don't think they charge a commission, though I can't be sure--as I understand it, they just make money by having you use your Visa card for the purchase. In this article[1] they tried to take it to an extreme, but frankly most of those tasks would be better for Fancy Hands.

Since I also have Fancy Hands and FH is better quality, I use Visa for tasks that are low priority, or I can afford to have redone a couple of times, or are very simple. Good examples of these include: not immediate restaurant reservations, basic research of all kind (anything from 'find me a place in NYC that sells Petrus Aged Pale Ale in bulk' to 'get me a list of all the language classes that NYU has for non-students with dates, times, location, and price' to 'how do I mail a letter to Spain'). Mostly generic things. Fancy Hands is what I use for specific, personal, or high-touch items--"Cancel my AT&T account," "Get PayPal on the phone again and again until they fix my account," "Please call this number in the UK during their morning and get this information so I don't have to stay up late," etc.

[1]: http://fourhourworkweek.com/2010/05/01/credit-card-concierge...


That was the first thing that popped into my head. I have been thinking of rejoining fancyhands as I get busier.


unless i'm missing something, the idea is to hand over all of your personal information to two college kids with internet access?


Said college educated. First glance I thought the same thing too


>artificial intelligence and futuristic tech.

That basically broke my BS alarm.


Hey, I was working on a project similar to that a while ago that interfaced with google voice! However Google voices api was deprecated and my project never really got any further. Here's a link to the git hub repo https://github.com/janbro/J.A.R.V.I.S

Edit: Scratch that, my project was nothing like their's, my JARVIS was meant to be a fully autonomous sms personal assistant. Apparently this is a service which assigns you two human people you text for things who are "aided" by artificial intelligence and futuristic tech, whatever that implies. Their FAQ page says a bit more about the service, although not enough for me to want to subscribe to it


Are you aware of any other services that can be used to integrate a bot with real time text message communications?


Yep, check out the Cyman Digital Butler for Android. With the paid version, when you get an incoming text you can set it so he reads out who sent it, then you can say "read message", or "reply" and then your reply. You can also initiate messages, and there are helpful features to help with complex names. Like if you said "my wife is called ...". You could then later say "send a text to my wife". I worked pretty hard on it!


If your asking if their's another service like Google voice, I'm not sure. I haven't done much more digging on text message bot integration, although you could make an irc bot which would operate on the same principle, just not through text, and not as private.


Not sure whether this will meet your need but i found voxeo[0] pretty good for voice related stuff. I had to discontinue my product idea due to the fact that they don't have instantaneous voice to text which was bummer for my project [0] www.voxeo.com


Grrr this isn't Jarvis.

I've been running Sphinx-pocket (speech recognition engine) on an old pentium4 PC in my garage. That's a way better approximation of Jarvis, even if I do say so myself...

Actually if anyone knows of a way to lookout for the word Jarvis without hogging the CPU...? The current setup uses a button (connected to an arduino running firmata to give the old PC some gpio pins) which is monitored by a daemon and triggers sphinx on a debounced button-up event.

I'd like some low CPU way to find potential candidate input from the mic that sounds like it could be an instruction worth passing to sphinx?


There was a project I saw on HN that did that using an RPI. I think this might be it: http://www.raspberrypi.org/meet-jasper-open-source-voice-com...

It's all open source, so hopefully you can find what you're looking for.


Great find. They're using pocket sphinx as well I see.

Their mic module listens for 1 second, establishes a threshold volume, then for the next 9 seconds it listens for a disturbance above a weighted threshold. They're sampling at a reasonably high rate.

Either way I'm going to try their code and see the performance. The approach should be faster than mine but I didn't see how they handle the edge case of a command coming in over the 10 sec boundary when it restarts listening.


I don't have the answer for you but if you'd like a real-world PoC the Kinect does exactly what you're describing only with the word "Xbox." Does it quite well too I might add...


I made a Cyman Digital Butler which runs on Android devices and on computers through a Chrome app. I feel it's a better Jarvis approximation too. On both of them, Cyman can listen out for whichever nickname you choose. (Of course, the nickname I gave was Jarvis for mine ;) ) I find using the Chrome APIs or Android APIs quite good if you were interested. I haven't really looked at sphinx


Okay man, we get it! There is absolutely no need to spam every single thread discussion on this article with shameless plugging about your "Cyman Digital Assistant".

Comments by ugoano:

Comment 1: I'd strongly advise you to check out the Cyman System website. The Cyman Digital Butler has a few video demonstrations, and I built it with Jarvis' personality in mind. Might be what you're looking for?

Comment 2: I built an digital butler app inspired by Iron Man's JARVIS assistant called CyMan. We've got a great community and dedicated helpers in our Google+ community. Feedback on that community has been excellent! If you ever wanted to give it a try...

Comment 3: My assistant (Cyman Digital Butler) has a list of example commands you can try, gives an idea of the abilities Cyman has. So you could say "open the ... website" to get the url for example. Although we've got a great community to help, we're in the middle of creating a wiki. Perhaps the Jarvis you're using needs a wiki too.

Comment 4: I made a Cyman Digital Butler which runs on Android devices and on computers through a Chrome app. I feel it's a better Jarvis approximation too. On both of them, Cyman can listen out for whichever nickname you choose. (Of course, the nickname I gave was Jarvis for mine ;) ) I find using the Chrome APIs or Android APIs quite good if you were interested. I haven't really looked at sphinx

Comment 5: Yep, check out the Cyman Digital Butler for Android. With the paid version, when you get an incoming text you can set it so he reads out who sent it, then you can say "read message", or "reply" and then your reply. You can also initiate messages, and there are helpful features to help with complex names. Like if you said "my wife is called ...". You could then later say "send a text to my wife". I worked pretty hard on it!


Sorry dude, got a little over excited when I see particular issues that I feel I (hopefully) have addressed, so make sure those particular people know of my attempt. If I was just spamming, I'd go for all of 'em :)


Interesting. How can this truly be unlimited (albeit 15 minute) tasks?


Jarvis, spend 14 minutes coding this project. And take good notes, I may need you on this again later.


The same way there can be unlimited bandwidth? It just gets real slow if you turn out to be an "abuser"-:)


Small typo on step 2 of the signup:

"We collect your card now though, in case you want us to purchase things durring the trial." (should be "during" instead of "durring")


I signed up last night. I figure that , I signed up. Haven't been thrilled with service yet. I asked them to set up some calendar appointments with my team and I already set up all permissions, etc... 14 hours later, no appointments on calendar.

They definitely need more information on their site and some thinking through process, etc.

To be fair, I'm guessing they got inundated after this went up on YC/TC. And the service looks promising. Here's hoping to some smooth sailing


Makes me wish Flow Concierge was never discontinued.

http://www.getflow.com/blog/2012/05/introducing-flow-concier...


I'd be interested to know what's their business model for the months to come. like is it even scalable? but I love the idea of disrupting the realm of personal secretary.


Wonder how they're interfacing with iMessage. As far as I know the binaries are heavily obfuscated and have no public interfaces.

I'm guessing the name is based on Jeeves?


Jarvis is Iron Man/Tony Stark's computer assistant in the movies.


Jarvis is the long-time butler of the Avengers mansion. That's where they took the name for the movie.


That's embarrassing, I know characters from books in the 1800s but not movies in the last decade.


like who from the books?


Well, Jeeves. The gentleman's personal gentleman.

https://en.wikipedia.org/wiki/Jeeves


I don't see why they would need to interface with iMessage at all. Perhaps I'm missing something, but it just seems like they're responding to text messages sent to a particular number.


I think iMessage is a smart choice. it lets their customers communicate via a Mac, iPod, or iPad in addition to an iPhone. Text messages also cost me money, so not having to worry about that is another selling point.

It also makes scaling their side of the system easier. Instead of having some sort of CRM/SMS gateway system, they can just hook however many Macs they want to to their account, and all messages sent to it will get copied to all of them. This makes it easy to hand off a communication session between agents in a way that's transparent to the users of the service.

Of course, the downside is that it's an Apple-only protocol, so they will be prematurely excluding users of non-Apple devices, but maybe that's a tradeoff that the company's willing to make at this time.


the screenshot they put on the homepage suggests that it does interface with iMessage?


AppleScript can interface with iMessage, so it's certainly possible to magic up an interface.


The screenshots I see just show a regular text message conversation.


Blue bubbles are for iMessage, green is for SMS.

It also says "iMessage" at the top of the conversation.


Good point. I didn't notice this distinction (I'm not current an iOS user). Although they claim you're communicating with humans, so they could potentially just be using iMessage at their current scale. Or the screenshots could be from a simple fully manual demo.


Maybe the concierge is using a Mac, or an ipad with Bluetooth keyboard, or a Hackintosh?

Couldn't you script iMessage on OS X with AppleScript, and thereby allow multiple people to interact with the same instance?


They're not. "Jarvis is a tech enabled person. You will be assigned two US based, college educated personal assistants..."


The service sort of implies otherwise. If it's a phone number they'll get texts and iMessages. If they're using the feature that forwards iMessages to an AppleID the SIM must be in an iPhone, and I hardly think they'd reply to every SMS from a single iPhone.


The easier answer is that the screenshots are faked and not actually sent through the service; they probably just renamed a real contact who happened to have an iPhone 'Jarvis' for the purpose of making screenshots.

All the screenshots have timestamps a few minutes apart showing a plugged-in iPhone with an increasing charge over the course of a few minutes.


iMessage (well, Messages.app) is scriptable on the Mac via AppleScript (see https://github.com/lazerwalker/hubot-imessage). Figuring out how to scale it might be tricky, but for the time being they could be using some Mac scripts to intercept the iMessages, shoot them off to the appropriate assistant via some other interface, and then send the resulting iMessages back through the same Mac interface.


it's not hard to set up a proxy and just sniff the traffic to reconstruct the protocol


It must be kind of hard, because nobody has reverse engineered it enough to build a reimplementation of the client.


Someone has, actually [1]. Dozens of people also seem to have at least partial implementations.

[1] http://imessageonandroid.com/


Really? Do you know what you linked to?

Have you actually gotten an APK from that site?

It looks like a 100% fake site that takes you through never ending scam surveys and sign ups.


That one could be wrong, I don't have an Android phone anymore, and stopped following pretty much all things Android. However I've personally used an iMessage app on Android a few years back. They definitely exist.


It's obfuscated and very complicated.

http://imfreedom.org/wiki/IMessage


Recently I was developing this kind of application on Android for my personal needs because 3G is recently launched in Pakistan and it is not affordable yet.


Not sure why was this down voted. I will clarify. By Android application I mean, a server side application which listen for incoming texts and respond to them via text msgs.


I use GVoice but it's rather limited. What do you use server side to listen to and respond to text messages?


I can't seem to validate '100$/month'. Do I have to sign up in order to see that info?


It shows up (with $99/month) after giving a name/email/phone number but before giving credit card info or completing sign-up.


Can someone tell us what kind of "artificial intelligence and futuristic tech" these guys have?


People. From the bottom of the page:

>Light Speed, Meet Human Intelligence // College-Educated & US-Based

It's college-educated people based in US answering SMS questions around the clock. No fancy-fancy, or even fancy of any sort.


Correction, 2 college educated people with laptops and access to google.


Jarvis is not a human being. Can we save that name for when true general AI comes up? : /


I actually find this super brilliant, however, a nitpick -->

'college educated - in the us' <-- that's all this kids can get is some bullshit VA job that pays dick/hr? lmfao "higher education" - what a scam 'college' is


Hi, we do UNLIMITED business tasks @ $69/month for entrepreneurs, managers, administrators & back office staff. :) https://www.timesvr.com/faqs


Funny, I been working on a very similar concept


I'd like to hear / learn more either as a future customer or if you need help with tech or bd, or just to throw around learnings.


As have I. Mine works with Google Voice. How do you integrate your bot with text messages?


I've always used http://www.twilio.com/




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: