We’re moving to Silicon Valley.

It’s happening.

We are joining Boost VC Tribe 10 batch with a new product for restaurants and coffee shops.

Thousands of startups were met and interviewed.

Only 5% of applications were accepted.

And we are one of them.

How did we get there?

It’s just work.

I do think it will bring a lot of value to this community because now the possibilities are almost endless.

The next interview is going to be with Amir Shevat, development relations director @ Slack 🔥

P.S. Anyone free for a coffee in SF? Let me know: @flreln 🙂

Hey botmakers, anyone would like to submit their bot for review in our #botoftheweek series?

The offer lasts for an hour.

Send me (@flreln) a message with your bot’s description, and I’ll choose the best one to write a post about 🙂

I haven’t found any compelling bots for today’s #botoftheweek.

But I want to share with you a trick we use to improve our bots onboarding & engagement 5x. It works 🙂

It’s based on a habit creation model called HOOK, described by Nir Eyal in his book “Hooked: How to Build Habit-Forming Products” (highly recommend to read).

Let’s say we are building a community manager bot, which helps to engage community members in Slack.

To implement the HOOK model, we need to activate a habit-forming process, which consists of 4 steps:

1. Trigger

The trigger is an activator of a behavior.

It could be either internal or external, and usually, in bots, internal triggers (such as boredom or curiosity) are activated by external triggers (such as a direct message from the bot).

The goal of the trigger is to drive the user to take action using the product.

In our case, the trigger is hidden in an onboarding message:

“Greetings, @flreln! Welcome to BotCube, the top Slack community for bot designers.”

”My name is Botty, and I’m a bot, designed to help show you around and get the most out of your membership.”

”Say “help” to get started or “tip” to get today’s tip about bots.”

The last message here is an example of a good trigger because it has 2 key things:
- Shows that the reward is variable (“today’s tip”);
- Is actionable and provides the user with a choice.

2. Action

The action is a simple thing user needs to do to get a reward. Like hitting “Show me” button or opening a notification.

Once the user has done the intended action (in this case, typing the “tip”), he’s dazzled by what he sees next.

3. Variable reward

The reward is the realization of the value from the action.

Feedback loops are all around us, but predictable ones don’t create desire and don’t surge the level of dopamine in the brain.

In our case, the reward is a random useful tip about bots that user gets after he typed “tip” message to the bot.

Variability and reduced accessibility here multiplies the effect, creating a frenzied hunting state, activating the parts associated with wanting and desire.

4. Investment

The last phase of the HOOK is where the user is asked to do a bit of work.

Here we can benefit from the reward we just gave him and ask him to take action that makes the service better with use and generates future triggering opportunities.

In our community bot example, we can promote a survey or explain any other feature we want the user to proceed to.

These investments can be leveraged to make the trigger more engaging, the action easier, and the reward more exciting with every pass through the HOOK loop.

That’s it.

I highly recommend you to implement this approach in your bots and just see what happens next.

You will be surprised :)

How could I miss it??

A friend of mine Stefan published an incredible list of best AI cheat Sheets (Neural Networks, Machine Learning, Deep Learning & Big Data) earlier this month.

It’s not directly related to bots, but I do recommend you to have a look if you’re going to build something that includes machine learning.

Enjoy your evening!

https://becominghuman.ai/cheat-sheets-for-ai-neural-networks-machine-learning-deep-learning-big-data-678c51b4b463

Do you remember that Black Mirror episode which tells the story of a young woman whose boyfriend is killed in a car accident? Later on she discovered that technology now allows her to communicate with an artificial intelligence imitating him.

Seems like it’s a reality now 😲

Video below is a story behind Replika AI. Replika is a chatbot that creates a digital representation of you.

It's strange and fascinating - but the story behind it is even better.

Eugenia Kuyda’s best friend died in 2015. Using a chatbot structure she developed, she entered their messaging history into a Google-built neural network, creating a bot she could interact with. It was the earliest version of Replika, a bot that, as you interact with it, turns into a digital representation of you.

This video makes me think about one interesting point nowadays - what it means to be human in a world increasingly filled with robots 🤖

https://www.youtube.com/watch?v=yQGqMVuAk04

Hey guys,

I have HUGE news.

We’re launching Bot Designers, Developers & Marketers community on Facebook.

This group is to share information, tactics, and strategies regarding creating bots.

No questions, no promotional content, no links to websites or email forms.

Only valuable content: interviews with industry leaders, 2-3 how-to materials every week, guides, bot reviews, and daily posts with tips & tricks.

Whether you're a designer, marketer, developer, or founder, this group is for you to learn, share, and grow.

Why create this group?
1. Create the one-stop resource for bot building
2. Provide a valuable opportunity to learn from industry experts
3. Giving back to people is super important

You can only succeed if you help others succeed, and that’s what we want to live by in this group.

I would love to see you there, right in the first members.

You can join here: https://www.facebook.com/groups/botdesigners/

There was no news from Facebook Messenger Platform team from April.

But here it is - Messenger Platform 2.1 🚀

Two main features are:

1. Built-in NLP support.
Starting today you can directly connect your Facebook bot to Wit.ai (startup Facebook bought 1.5 years ago) without coding.

It'll automatically detect meaning and information in the text of messages that a user sends before it even gets passed to the bot.

Isn't that cool?

Nice move by Facebook to popularize Wit.ai as an NLP provider - make it native.

2. Handover protocol.
This protocol enables businesses to create multiple experiences within a single bot (connect multiple apps to one Facebook page).

There were a lot of dreams about one main BotFather that can handle all the requests and redirect you to different bots.

Seems like we're a step closer to such a world right now.

I genuinely think that a "Handover protocol" feature is opening a huge amount of opportunities that were impossible before.

Already thinking about applying it to the bots we've built.

You should think too! 💪🏻

https://messenger.fb.com/blog/messenger-platform-2-1-brings-new-tools-to-enrich-conversations/

I suppose that most of you do know Statsbot.

For those who don’t know - it’s a Slack bot that allows you to get reports from data sources like Google Analytics, Mixpanel and others using natural language.

But here is a new big player here - Google.

They've just started rolling out a new Google Analytics feature - "ask a question".

The aim is to make it much easier for people throughout an organization to be able to get the information they need without having to rely on data analysts or needing a deep familiarity with the Google Analytics interface.

So, you can just go to your Google Analytics panel and ask: "What's my revenue last month on mobile vs. desktop?".

Yea, that simple.

Working example GIF: https://goo.gl/oBBpjM

More about the update: http://marketingland.com/google-analytics-ask-questions-intelligence-219961

Google launches a great program for AI startups!!

These startups usually lack experience and data to build a good prototype, that’s basically what Google wants to fill up with this program.

They are aiming to address these needs head-on with specialized data sets, simulation tools, and prototyping assistance, which is super great for an early stage startup.

Another good point for already established companies is that startups accepted will have access to Google talent, including engineers, IP experts, and product specialists.

The Launchpad Studio will be based in San Francisco, with additional operations in Tel Aviv and New York City. Eventually Toronto, London, Bangalore, and Singapore will host events locally for AI founders.

If you’re building something relevant for this program, I’d highly recommend applying (no matter are you an early stage startup or an established company).

Apply here: https://events.withgoogle.com/ai-studio/registrations/new/details/#current-step

More on TechCrunch: https://techcrunch.com/2017/07/26/ai-studio/

Designing a great bot personality is hard.

Designing a great bot personality for multiple channels is even harder.

We all want to do it once and use it across all the interfaces - voice, text and even rich UI with buttons and stuff.

But that won't work out.

You need to think about each channel separately.

For the very best conversational UX you have to do some work tailoring your content and personality to each situation .

For example in a voice-only interaction, brevity and clarity are much more important than personality. Functionality goes first.

In text interactions, personality becomes especially key when the conversation goes wrong.

The article below covers a lot of important points that can help you design your bot personality to fulfill user needs on every channel or platform.

https://chatbotsmagazine.com/responsive-personality-design-in-chatbots-a96b1c2373ba

Travel is a great market for chatbots, and here’s the proof:

SnapTravel, a bot that lets users find and book rooms via SMS texts and Facebook Messenger, just raised $8 million.

Unlike a lot of chatbots on the market today, SnapTravel isn’t a fully automated solution.

The bot handles onboarding and answers users’ simpler questions, and human agents are always ready to help with the tough ones.

For example, if you booked a non-refundable hotel but your flight gets cancelled due to weather, an agent would respond to your questions and try to help, rather than leaving you to deal with an unsympathetic bot.

Here are top 9 insights from their journey:

1. Chatbots are great for onboarding a customer. A user can quickly engage with a chatbot as it is a simple UI they are familiar with. Chatbots are also great for answering simple questions.

2. Human Agents are necessary for tricky situations. If you booked a non-refundable hotel but your flight just got cancelled due to a storm, you want to chat with a live human agent — not a chat bot! At SnapTravel — our live human agents are available 24/7.

3. A Webview (mobile web) is essential for most hotel searches. If you are a business traveler who has booked with us many times before and you make a request — we could just send you a booking form for the perfect hotel.

4. Building on top of Messenger means that we have access to a new and untapped source of distribution (discover, drawer…).

5. It means that we can engage a user without forcing them to download a new app.

6. It means we can leverage sponsored messages — one of the most untapped ad units on the platform.

7. If a user makes it to the booking form but doesn’t complete the transaction, we can retarget that user directly in their messaging inbox as opposed to trying to follow then around the web with display ads. We have seen a 30%+ jump in conversion rate with effective Messenger retargeting techniques.

8. Continuous A/B Testing is essential — The leader in the hotel booking space (Booking.com) is well known for their rigorous A/B testing platform. Similar to Booking.com, we have invested in internal tools that allow us to run hundreds of tests in parallel, anything from messaging frequency, tone and length to web ui fonts, colors and interactions. Due to our relentless focus on data driven experiments, we’ve seen a 2x higher conversion on mobile compared to traditional bookings websites.

9. Delight the customer — One of the benefits of working with a traditional travel agent is the personal service and attention you get. However, no one has the time anymore to walk into a travel agency, or even pick up the phone to call. At SnapTravel, we have invested significant resources in delighting our customers — providing 24/7 human agents/customer service over messaging.

https://medium.com/@hussein_fazal/snaptravel-8m-series-a-raise-conversational-commerce-77c389fa4edf

Amazon is about to release a new chat app that could be the WeChat of North America.

You may think “Really!? Yet another messaging app?”

However, the idea is not that far-fetched for several reasons:

1. Amazon has already been building messaging services, both for enterprises and consumers.

2. Amazon has already started to dip its toes into consumer communications, namely with Alexa Calling.

The free calling and messaging service, Alexa calling, was launched in May.

It’s not so hard to get that Amazon was from the start expanding the communications experience outside of the Echo-device ecosystem: people could also use their Alexa apps — on smartphones and other hardware — to communicate with each other.

And it makes a lot of sense that Amazon would develop a more focused, standalone messaging experience as a progression of that.

Anytime hopes to be an all-in-one feature rich service that could even rival social networks.

Some of the possible features are:
- Food ordering in group chat
- Bill splitting
- Embedded gaming (remember they own Twitch)
- Video calls with filters
- Ability to make reservations
- It will have gifs!
- Video and audio calling
- Secure messaging and photo filters
- Support for video and photo filters
- Ability to @mention other users (just by name)

We’ll see how ecosystems like Alexa, Prime Video, and Anytime will bridge the consumer into the Amazon experience; that will be able to facilitate everything, not just shopping, anytime they want.

Interesting times :)

https://techcrunch.com/2017/07/14/amazon-messaging-app-anywww.tg-me.com/

My friend Mike, who’s the CEO and co-founder of ManyChat, is sharing a real value bomb!

They’re going to release ManyChat–Zapier integration very soon and now are giving EARLY ACCESS to this channel!

The initial integration will connect ManyChat to 750+ other apps and will enable use cases such as:

1. Ask a subscriber for an email inside the bot and add it to your email marketing service / CRM.
2. Ask a subscriber for their phone number inside the bot and add it to your CRM / send a Slack notification.
3. Ask for any text (feedback / suggestions / rating) and pass this text to Google Sheets / Slack / email.
4. Trigger an email sequence based on an action inside the bot (e.g. user pressed a button saying "Get the 7-day crash course" and you trigger an ActiveCampaign sequence to this user).
5. Register people to a webinar with 1 tap, if you already have their email in a Custom Field.
6. Move the lead in your CRM pipeline when he/she get's a certain tag in ManyChat.
7. Add a lead in your CRM when you get a new subscriber in ManyChat.
..and many many more.

Disclaimer: This is early access, so there is a chance things will break.

If you want to get an invite, send Mike your email here: m.me/mikedyang

Because we love you guys ❤️

Facebook is going huge on improving their AI assistant M.

They've just acquired AI assistant startup Ozlo to help build “compelling experiences within Messenger that are powered by artificial intelligence and machine learning”.

Ozlo initially focused on helping users sift through restaurant listings in a conversational format.

But they've eventually expanded to include not only restaurants but also weather forecasts and more local business services, like movie listings, leveraging popular APIs like Yelp.

As we can see guys from Ozlo were doing pretty the same thing Facebook AI assistant M is trying to build and became a pretty valuable asset.

So happy to see all of these acquisitions started emerging in a chatbots space 💪🏻

https://venturebeat.com/2017/07/31/facebook-acquires-ai-assistant-startup-ozlo/

Oh, sh*t. That’s the Hook Model!

Have you ever spotted a popular framework in action?

It’s a strange yet reassuring feeling!

Realising someone has implemented a strategy, textbook perfect. Nir Eyal would be proud.

Well that happened to me yesterday when I was reviewing Champ for our #botoftheweek series.

Champ has stormed to success in the last couple of months, processing over 25 million messages on Messenger, here's my review as to why.

Full review here, made by Matthew, one of our BDDM community admins:
https://docs.google.com/document/d/1E1FPlwEF4ZBFFe8NonVbiPPySM0rZne4E4fsggeV1pg/edit?usp=sharing

Smart speakers market is growing rapidly.

Google and Amazon have already sold millions of their smart speakers in the last year.

Xiaomi, Alibaba, Microsoft, Apple and many others have already launched their own smart speakers recently.

Facebook doesn't want to miss the opportunity.

Their research lab is reportedly working on a device about the size of a laptop for video calls, similar to Amazon Echo Show, and a separate smart speaker.

The devices could be available as soon as spring 2018.

Something big and completely new is coming.

Amazon Echo has only 15000 skills, Google Home has something around 400 and Microsoft Cortana has only about 70 apps available.

Can you imagine that App Store, for example, has only 70 apps available?

That's crazy.

Don't miss the train, it's going super fast.

https://venturebeat.com/2017/08/01/facebook-is-reportedly-developing-a-video-chat-device-and-smart-speaker/

When we were launching our first chatbot project I was building the whole system from scratch manually.

I was coding about 12 hours a day.

Our only competitors were doing the very same thing using Chatfuel.

"They won't be able to scale. If they need to add some custom functionality - they're dead" - I thought.

That project hasn't worked out.

Our competitor's product is still alive and acquiring top clients.

You don't always need a custom solution, you can start from using bot building platforms to build the prototype or MVP.

Here is the list of about 50 bot building platforms with the detailed comparison.

Here you go:

https://goo.gl/Z4SXuo

I used to hate Skype even before it was acquired by Microsoft.

It was buggy, ugly, old-fashioned and completely unreliable.

We've moved our internal team communication to Slack about 2 years ago.

But I see a lot of positive changes there for both common users and chatbot developers.

Microfost is really pushing Skype in chatbot direction with new releases of Microsoft Bot Framework, web chat plugin, releasing a separate tab for chat extensions and plenty of other stuff.

And right now they're adding a new feature there - payments.

The feature is pretty similar to what Facebook Messenger released a couple of months ago together with PayPal - you can easily split any bill across a number of people via chat extension.

Skype’s choice to partner with PayPal was a smart one, as it’s an established name many people have already signed up for.

Considering many business sessions happen via Skype too – like tutoring or singing lessons – it could make it easier to collect payments from clients.

I hope Microsoft will continue pushing Skype in this direction so I can start using it for some of the business communications again 🙏🏻

https://thenextweb.com/apps/2017/08/02/skype-now-lets-send-money-via-paypal/