This whole 'bot renaissance makes me think that somewhere 1000s of old IRC Channel admins are raising their hands in despair, and quickly shuffling through countless CD-Rs of old TCL and Perl code to see if they have something worth re-releasing.
Granted, the natural language stuff is way better than it was, but the current bot platforms that are springing up everywhere are nothing more than a language parser and an empty ruleset DB that you as the customer have to spend endless hours populating.
It seems a very unfinished solution right now; We're going to see a flood of sub-par bots* on the market that are going to cause the end users to disengage within months.
(*I'm sure motion.ai's bots are very good, and I am not singling them out here, just commenting on the whole bot thing in general)
My main concern is that this whole situation will not include only chat bots but it will also trigger a demand for crawling bots which could put a lot of stress on web sites. Then sites could adopt aggressive captcha schemes which will hammer usability and make the life of legitimate crawlers a lot more difficult. It’s a lose-lose situation , until the industry comes into its senses and realize this whole thing is a fad.
One solution to this could simply be improving our ability to handle traffic. Bots aren't doing anything bad (under normal condition), they just represent a portion of your view traffic that you don't inherently want. An aggressive user, of sorts.
With that said, we have more cycles these days, better web servers, and inherently more traffic capacity than we did 10 years ago.
I have a feeling bot traffic will grow faster than our traffic-handling-tech, but regardless, i think this is just going to be the new standard. Especially since bot traffic could represent UX requirements for potential users.
I don't understand the connection. What do chatbots have to do with scrapers? Are you saying that the chatbots will need to use scrapers to harvest the information they present to users? Might be true in some cases, I guess.
Should be on the home page, perhaps even above the fold already asking you a question with a side-by-side of the flowchart components being activated/highlighted.
That's a super good UI suggestion, thanks. We just launched our public beta about 6 hours ago and have been thrilled with the response and very unique bots being created (particularly when people leverage IFTTT and Zapier too) ... we'll definitely be surfacing as many cool uses as possible ASAP :)
From an NLP standpoint, what is the comparison between this and other NLP products? Namely WitAi, etc?
I'm writing a bot framework (code, not web service), and trying to generalize the process so the the developer can switch between different NLP providers with ease (to avoid lockin.. as much as possible atleast).
So, comparing to Witai atlast, is your NLP better? Is it more flexible somehow? etc, thoughts?
Our value-add here would be providing the pipelines for one-click deployment (SMS, Facebook, Slack etc) and structuring of conversation flow, in addition to things like post-deployment analytics and insights on how users are talking to your bot. In terms of NLP, we actually take an agnostic approach to whether users want to use Wit.ai (or api.ai etc) in conjunction with our service, or use our NLP offerings. Thanks for the question!
Most of what I've seen is in English but there are some second-level services popping up that promise an API for chatbot translation, e.g. Cyrano (http://cyrano.unbabel.com/).
I remember the time when kids considered bots cool, and I had no idea about programming then, nor did I understand how bots were usefull..
But the problem is I dont understand not even today who would be interested to interact with a bot?
And how is this service better than employing a programmer ?
Some examples would be useful.
The platform is only hours old in terms of public availability, but here's a pretty cool customer service demo bot that scans a knowledgebase first, but can create a helpdesk ticket if human assistance is required: https://www.youtube.com/watch?v=e9IMsJW4ca0
Maybe I'm missing something, but can't a normal website do all that?
Or is the idea that you can feed message-only channels like email and Facebook messanger to it? But then you could just say "We don't really provide email support; see our website here: ..."
Kind of feels like using telephone menus when you don't have to.
(Nothing against your platform though - it looks very well done!)
Bots are broken as a platform. No advertising, no payments system, developers have to pay for hosting, there's no discovery mechanism because the messengers don't have bot stores... Why would developers be interested?
I don't understand the discovery comment. I don't discover Bob the copier salesman. I decide I need a copier, search for copier retailers and then give Bob a call. Won't bots work the same way?
You don't have to discover how to call Bob, which is analogous to finding the bot. But you may need to discover what jargon to use with Bob to make sure you get what you want.
Depends on which vision wins out. The phone handset vendors certainly have a leg up on being the "discovery" part, deciding which copier retailers you get exposed to.
I'm not trying to convince you otherwise, but I honestly believe you're missing the point. Bots will triumph not because of technical reasons, but rather because they are what people want. If you let me a self-plug, here's a detailed explanation I wrote a few days ago, I'd love to know what you think after reading it: http://cfenollosa.com/blog/bots-lack-metaphors-and-that-is-t...
As I understand it, bots have been really successful in Asia. The messaging platforms did add payment ability for the bots, and it doesn't require users to install a new app.
There also are bot stores on some messaging apps. I know Skype just added one. And facebook and telegram were giving out grants for users to develop interesting bots on their platforms.
I got an idea the last time I saw this discussed, which probably will make some present HN people rich:
Bots which follow your own conversations/emails/etc, read the comments from other people -- and in another window suggest jokes, information, links etc based on parsed language. Just click to add the suggested answer.
You could install modules to see suggestions from e.g. a stand up comedian, some experts for different topics or just the wikipedia pages for all mentioned non-trivial concepts, etc.
A bot could e.g. help with fixing grammar errors.
Everyone can be their own Woody Allen -- or get help to write as a depressive persona. Or as a Romeo. :-)
Granted, the natural language stuff is way better than it was, but the current bot platforms that are springing up everywhere are nothing more than a language parser and an empty ruleset DB that you as the customer have to spend endless hours populating.
It seems a very unfinished solution right now; We're going to see a flood of sub-par bots* on the market that are going to cause the end users to disengage within months.
(*I'm sure motion.ai's bots are very good, and I am not singling them out here, just commenting on the whole bot thing in general)