Congrats Hu:toma team! So much progress made since last year StartupBootcamp demo day :)
the bot landscape has evolved so much in the same time.
I have added you to the Made in Barcelona collection https://www.producthunt.com/@pic...
@picsoung the product and team have developed a way since last year. Now looking forward to getting some feedback from Beta users as we onboard them on to the platform over the coming weeks. Do you see any use cases where you think Hu:toma's method of uploading text to train a bot trumps a chatfuel style approach and vice-versa?
Ive been using api.ai to configure a chatbot for our coworking centre BASE10, where our members can ask (in Slack) various questions eg 'how do I fix the coffee machine, how do I book the event room, etc'. I had three questions, how would using Hutoma be better than api.ai (aside from the fact their support is non existent so would easily cause me to swap for just this one issue). Api.ai seems to be free (at least at the moment), what is your pricing model? And finally, one of the features I desperately want is a way for when users interactive with the bot, instead of giving an answer to things the bot has no response stored for, instead a human (admin/designated person on slack) could step in and answer on 'behalf' of the bot, at the same time training it for next time. A sort of 80% bot and 20% human (when the bot doesnt have a good reply). But giving this flow to users means they have 100% faith in the bot and it becomes less of a gimmick and more of a reliable tool. Is such a solution possible with Hutoma?
@jasondainter Hi Jason thanks for the reply. Well, we hope to provide better support so I hope to see you soon on our platform. In general we can score better accuracy compared to API.AI and Watson. We wont charge devs to use our API and only apply a rev sharing if a bot is sold. We are already handling an handoff feature with one of our enterprise client but we dont have it in our dev platform. However I just added it to our next sprint planning and I promise we will make it avaiable. I think your point is very valid. Thanks a lot!
This is an interesting approach — an independent bot store build on a single platform:
Hu:toma is a marketplace and proprietary deep learning technology for chatbots. Developers can create and train Siri-like chat interfaces using only text samples with no coding required. The "Bot Store" allows them to find customers and monetize their creations. Companies can leverage the power of readymade neural networks to build and scale their business processes efficiently, empowering their team to work on higher value tasks. End users can interact with the chatbots via Facebook Messenger, Slack or a mobile / Web application.
My question is: how will regular users find and interact with bots built on the Hu:toma platform?
@chrismessina Chris many thanks for posting out us! With Hutoma we want to provide the tools and an ecosystem where AI creators can get together to build, share and monetise their bots. Developers can mix and match existing bots from our store and can build on top of that instead of starting from scratch.
Once a bot is created, you can test it within our system (see this video for example:) or you can export it to one of the many messenger platforms such as Facebook Messenger or Slack. You can also just embed the bot into a web page. Regular end users will be able to interact with the AI directly from those platforms. If for example, you have exported the bot to Facebook, you can search for it within the messenger platform and can interact directly from there. We will be also exposing our store to search engines so you will hopefully be able to find what you need from Google or peraphs our own search functionality within the platform. Developers can also have access to the raw rest API that connects to the AI so if they want they could build a mobile app on top of that.
Many thanks again for posting about us and look fwd to be part of this community!
@chrismessina I think like many bot platforms we are reliant on the adoption and promotion of bots by chat platforms. By providing integrations to the channels end users are active within we think that familiarity and adoption by end users will increase. 1/ How do you see bot platforms facilitating this process faster?
2/ Our free-to-use subscription could come with branding, meaning every time a user interacts with a Hu:toma powered bot in Facebook messenger they see our name. I could see such branding having a negative impact upon UX, and because of this do you think it's something bot creators would pay to remove?
This is interesting @maurizio_cibelli
I've been playing with api.ai and they have a store for purchasing Intents to use... Thought the prices was way too high for someone just playing around with various bots and not intending to commercialize them. Would be interested to see if things that I build could be used as the base data for others and maybe generate a revenue stream? What the split with developers?
@helloduane hi Duane - thanks for commenting. Yes our goal is to support AI creators with free tools and monetise only if the ai owner monetises. We believe its important to create an ecosystem where everyone can experiment without worrying about paying. Would love to support you with your api.ai bot. Please keep in touch
@helloduane@maurizio_cibelli I've been using api.ai also. So far its been free for me (they seem to be very mysterious though about their pricing). What pricing did you pay?
@jasondainter@helloduane@maurizio_cibelli We allow unlimited testing of the bot from within our platform. Once you integrate to an external platform, FB Messenger, or your website we will charge a fraction of a $ fee per API call and offer subscription packages that contain X number of API calls.
For bot creators we offer an App store like revenue share model for every bot that is sold, split isn't concrete just yet but we're looking to Apple App store / Google Play as examples.
Really interesting!
Curious about what the Botstore has to offer, looks like it's easy to deploy. This in combination with the Neural Conversation Models is unique, haven't seen that approach before. How much data do you need from users to benefit from this?
@jimgroenen Thanks Jim! It's indeed very easy to deploy. The more data that's fed to train the bot, the better but it really depends on your business and needs. For an SME with limited interaction with users, there's not much of data needed to start having meaningful conversations.
@matthew_clementson Nice to see that the project hasn't died! This is a really new approach, really wanna see how it turns out. Love that there is a "small talk core" included with each bot,.. Just wondering, what happened to HER? :)
@ianissoawesome Hey Ian, thanks again for spotting Hutoma back when we were a Kickstarter project. During Startupbootcamp we decided to pivot towards a platform / marketplace model and leave the individual bot creation to you guys. Domain experts.
Fancy picking up the challenge and building HER on Hutoma?
Thanks a lot Chris!! Matthew here, part of the Hu:toma team responsible for growth.
I met back Maurizio and Andrea back when they were being accelerated by Startupbootcamp, where they shaped an idea using AI to ingest movie scripts and teach Maurizio’s son english into a commercially viable AI bot platform.
I’ll be here, along with a couple of colleagues to answer any questions you might have about the team, product, or market for AI bots. We’re here at 4YFN, part of Mobile World Congress - booths K3.10 and E2.5 so if you’re here too, lets meet up.
This looks awesome! I currently have experience with Chatfuel combined with api.ai. Will Hu:toma be able to make decision trees like Chatfuel and have a training module like api.ai?
@nimmegeersstef hi stef -- our tech works a bit differently from chatfuel and we are building a module to import api.ai. However if you need any help would love to get in touch directly.
Really interesting and congrats on the launch. We have been playing around with some of the bot frameworks as we scale our veterinary triage and patient intake service. This looks like it may be an interesting bridge between the various vet-tech operated hotlines (for which there are 10,000s of previous patient consultants) and the digital vet assistant of tomorrow. Looking forward to playing around!
Awesome staff! Totally recommend it. Specially for SaaS businesses. Was lucky to be one of the first adapters. The technology is great and mainly it works. As easy as setting up a Wordpress
Typeform