Building bots for services (Year One)
Today marks the first year since launching Nimblestack. A product shop that has one purpose, Build Fast. We have created our process of building software products in 24 hour sprints by attending over a hundred hackathons over the last 6 years. We have built tools that help us go faster with smaller teams and we’ve tested or used most of the available chat tools to build all kinds of bots. This is our journey.
In late 2015, at Giants Stadium somewhere between August and September, we (Aaron Franco and Gabriel Ortiz) built a concept for an Internet of Things Hackathon. It was called “Uber for Slack” (No, we’re not these guys) - Ben Parr was a judge (cool dude BTW). He seemed to like it. But was it a connected concept? For the judges, not really. We felt that it was the “internetting” of a “thing”, but the jury disagreed and we won nothing.
Some people thought it was interesting, but most people weren’t sure there was any value in a text version of a popular app. No one could see outside of the Uber/Slack box. At the time, no one believed that A.I. was a real solution to a customer problem. Some saw it as a moonshot solution to an unseen problem.
So, despite the far-fetched and the out of bounds, we took it on the road anyway! We showed a few hundred people “Uber for Slack” at events in San Francisco and New York City.
It didn’t make any sense! It was simple! but most importantly, it was cool. That didn’t stop the obvious questions “Why wouldn’t I just use the app?”, “How does this make Uber better?”.
Since we couldn’t get people to think outside of the Uber/Slack box. We decided to build something that would take a look past the product and into the future.
We started looking at the possibilities of combining A.I. with a service on a chat client. We thought that would be really interesting. And so, we came up with the concept for NimbleChat, followed by some preliminary testing and then set out from there.
In the beginning, NimbleChat had two services attached. Uber and Delivery.com. It didn’t really have a connection to Delivery.com, but that had more to do with our misinterpretation of the API documentation than anything else. In the beginning, it was mostly conceptual, but the demo was FIERCE and best of all, it sparked imagination!
Unknown to us, six months prior, Chris Messina coined the term “Conversational Commerce”. Now remember, this was waaaay in the past. Before the Amazon Echo, Google Allo and real-life self-driving cars went mainstream.
Which is funny to us because Uber on facebook messenger was probably only a notion at this point. This was a time when bots were a cool idea, but not widely regarded as a feasible idea. At the same time, OnDemand services were taking off big time and some startup teams were looking at messenger based ways of delivering services.
From the combination of these things and a sprinkling of NLU, NimbleChat was born! As an app! We built a way to ping coworkers over slack to find out their location. It was pretty cool, but no one understood how to use the app. So it was back to the drawing board!
Postmates for Slack
We also tried connecting Postmates to Slack. It wasn’t very good. We put it on the shelf for a while.
Techweek NYC ‘15
We took our idea to the streets with one main objective: find out if people understood the idea of an A.I. powered messenger based service. We got a banner for NimbleChat and stuck it on top of a table at Techweek NYC. It was well received as a concept and it was well-respected as a good idea. People certainly wanted it. The response from the public encouraged us to go further with it. Without Venture Capital, without Angel Investment, we set out to build a business upon hackathon wins, pure grit, client work and building messenger based bots. Little did we know, chatbots were on their way to becoming a “thing” on the internet.
What is a bot business?
…we were about to find out.
At Techweek NYC, we were approached by Sam Jain the CEO and Founder of Fareportal (makers of CheapOair) to build a version of our chatbot on top of CheapOair’s booking API with the addition of a Natural Language Understanding system. We all agreed that the future of chatbots was imminent and so we got to work on a chatbot for Fareportal. The biggest challenge would be in how people would react to the bot within a certain context. Challenges like these excited us, and so began our journey into building a chatbot for a corporate client.
As all corporate contracts go, it took a while to agree all of the terms. In the meantime, we did hackathons. And of course we focused on chatbots and their uses. Safe to say, it was a wild ride. Before we concluded the CheapOair bot, we got to work on all the possible bots. Our raison d’etre is building all the potential bots. For us, bots is LIFE! #datbotlife
We attended Launch Hackathon where we actually met Chris Messina and showed him TravelPack.club we “think” he liked it (also a cool dude BTW). In the end we won 3rd place prize for best API.ai implementation. It was a fun project. See the demo below.
BrewHacks (The beginning of the beginning)
BrewHacks was a hackathon hosted by Alley NYC and sponsored by ZX Ventures, I’m not completely sure what they do, but our best guess is that they are a consultancy office for AB-InBev’s new ventures. Using Delivery.com’s API to power the beer inventory for a Facebook messenger and a faux website that had an extensive beer menu, BrewBot showed the room what an robot selling beer on Facebook would look like. We worked hard to present something relevant using our knowledge of chatbots. In the end the judges were nice enough to select us for the $10,000 grand prize.
BrewBot (AKA Sir Veza) The loveable beer slinging bot!
Winning first prize was real validation for us. First off, many of the submissions were chatbot focused projects because Facebook had just released their messenger API and Amazon Echo also had just released an extended SDK. Off the back of this win, we took on two female developers to expand our team. We then set out on making BrewBot for really reals.
MailBot: “Emails go in… Summaries come out!”
MailBot was a Cisco Spark Bot that read your emails on Office 365 and summarised them using IBM Watson’s A.I. — it was a departure from our focus on ecommerce bots, but we think it was an idea worth exploring.
We won Three Partner Prizes at the TechCrunch Disrupt Hackathon. Now it’s Live on Facebook, but only works with GMAIL. Warning: It is a really early beta product, we can read your emails if we wanted to, we wouldn’t, but also we wouldn’t recommend using a sensitive email. Just make a new gmail account if you’re interested in testing it :)
“CheapObot” soft launch:
We started building CheapOair’s Facebook bot around the beginning of May 2016. In the end it took us a total of 52 days to complete a fully functioning chatbot that relied on CheapOair’s booking APIs. The functionality was limited to booking one way or return flights. It is currently offline whilst we figure out how to incorporate it after beta, but all in all it was a cool project. See an example of dashboard view of a real conversation with a real customer.
First big customer — Business goal achieved
We’re ready to build more bots for big brands. We’re also still doing hackathons, mainly building bots and new bot concepts. If you want a competent bot agency who loves to build fast, do not hesitate to get in touch.
TVbot is an A.I. bot that allows you to control the DirectTV set-top box with your voice. It was an awesome hack that won us two top prizes at AT&T’s Shape Hackathon.
Benjamin: The Budgetting Bot
On Benjamin, a user is able to create a login, and by linking their bank accounts to Benjamin, see a summary of their financial situation on a dashboard. A user is then able to add saving “goals” to their profile (items of interest they can search for on amazon.com), and given their current earnings/expenses, see how many days it would take them to purchase said item of interest. They are then able to select common expenditures, such as monthly Starbucks purchases or restaurant expenditures, and compare how many days it would take them to achieve their goal if that money were saved instead. Once the user sets their savings plan, our Bot reinforces the plan.
We will build more bots! We will support the bots we have. Do you need a bot built? Contact us here!
What is the quantum of what we’re doing? Simple, it’s the future of everything.
We believe that our discoveries will contribute to a greater software paradigm. Is our idea a game changer? That remains to be seen, according to a recent post by Arun Uday we’re betting on the WRONG THING.
Onward towards Conversational Commerce!
We believe in the growth NLU/NLP platforms and know that we can do better with canned responses when they are geared toward the user in a more thoughtful way. We didn’t stop when they said Uber over chat wasn’t a thing. We shall instead continue to build ecommerce and customer service chatbots for the foreseeable future. In opposition to the above, Intercom stated recently in their blog, most people are missing the real value in commerce over messaging, not us though :)
Doubling Down on NimbleChat
Right now, We’re taking our year of learnings into a v2 of NimbleChat. We’ll make that announcement in due course. Currently, we remain unfunded. We’ve been generating revenue since we started this idea one year ago. However, we think now we’re at a point where it makes sense to raise money, if you’re an investor and you’d like to talk to us, you can either message us via one of our facebook messenger bots or email me directly at: email@example.com