More than 40 enterprise bot startups, representatives of major messaging platforms, and interested developers and entrepreneurs gathered Monday to discuss the rapidly evolving world of bots.
Botness, led in part by our own Tom Hadfield and Ben Brown, was a chance to get like-minded people together who are excited about conceptualizing, creating, building, and deploying messaging applications and bots. More importantly, it’s also a time to start engaging people about the tools to build these things and the platforms for where they’ll live. At Message.io, one of our core missions is building an ecosystem — a giant, thriving place for these bots to live. We say, build it once, deploy to every platform.
Ripe for Disruption
For the first three days of SXSW, much of the talk about bots has been about how transformative they are. The ripple effect of bots and their ability to impact our work seemingly transcends industries and verticals, from banking to e-commerce.
From session to session, many developers have illustrated the leaps bots have been making, and it’s not unlike the evolution we’ve seen take shape from the onset of development triggered by the earliest days of the internet.
We have a belief that bots will change the way we interact with the world around us, much like graphics in the 80s, the web in the 90s, and mobile in the 2000s.
The transcendence powering bots ability to change is rooted in our human nature to communicate with each other — “the conversational workplace,” — a landscape that ranges across almost every imaginable vertical.
“We’re beginning to see this already in startups,” explained Tom Hadfield at the start of the Botness event. “Entrepreneurs run entire business in Slack. Messaging is the connective tissue between disparate teams, users, and data sources.”
Making waves in the enterprise is no small feat. Bot entrants here occupy a space worth $620 billion a year, worldwide.
Enter, the “Startup Idea Matrix.” With each new wave and s-curve, new entrepreneurial opportunity is created. For example: Salesforce displaced dynamics. Google displaced Exchange. The change creates new opportunity, and within each category, the opportunity for entrants into the multiple-billion dollar company club.
The enterprise space is huge, $620B a year worldwide.
Best Practices for Bot Design
We were fortunate to be joined by a panel of entrepreneurs looking to change the way we work with bots in the enterprise. Here’s some of their viewpoints, lessons, and advice for entrepreneurs to take into consideration when tackling that next bot venture.
CEO of Troops
Dan saw an opportunity for customer relationship management (CRM). Every company needs one, but the problem, he notes,“if it’s not in Salesforce, it didn’t happen.”
Worse yet, “if you don’t update your pipeline by Friday, you’ll be fired.”
Dan is changing that with what he calls, the “3 P’s” — personality personalization, and permissions.
- Personality: Notify the company/team when there’s an important change – a notification, but with “delightfulness,” such as using Giphy to inject some personality.
- Personalization: Enterprises and companies are different and unique: being able to customize info/dataset based on persona/use case.
- Permissions: User roles – in our dashboard control permissions with badges, features.
— Message.io (@message_io) March 13, 2017
Ottspott is a business phone bot designed for support and sales teams.
Luis has the ultimate buy-in, support from his team of 8, who were involved early in building the bot and use Ottspott for their own support. Insight can often come from non-tech people.
Some additional advice:
- Understand your bot’s audience
- For Ottspott, that looks like: buyers -> influencers, and then -> users
- Define usage & benefits for each user segment
- Not everyone has the same motivation
- Measure and refine constantly
- A lot of features we liked were not used
- Need to monitor what people are really using
- SMS was basic, then we had requests to deal with managing threads
- Use your bot as a customer loyalty channel
- From time to time, we can notify admins of new features
Lessons in Building Profitable Enterprise Bot Businesses
Dan doesn’t think of Donut as being in the bot business. Instead, he views his business as one helping teams to connect. The first step to achieving that goal and reaching some level of profitability is having a useful tool.
— Amir Shevat (@ashevat) March 13, 2017
“Stronger social connections across companies drives productivity and retention. Donut pairs up everyone in a channel and sends them to lunch,” Dan says.
Today, more than 8,000 companies are using Growbot, a bot that helps companies encourage and share accolades with colleagues in Slack. They do it with simple ways to tell employees they’re appreciated — keywords, emojis, and a light layer of gamification.
Here’s some of Growbot’s best practices:
- Admins and permissions
- Growbot is social, so everything you hear or see is usually positive
- As Growbot goes to more channels, there’s requests for more configuration
- Enable the admin to make changes for people
- In the early days, we did this ourselves, now we allow people to connect with slack admin directly to make configurations to the bot
- Make the bot easy to configure
- Every notification requires an opt-in for us to re-engage
- Opt-in before you start the relationship
- Be respectful of that relationship with your users
- Just as important
- Make it easy for admin but also the employee
- For example, our 10 minute recap, people being able to opt out
- Be able to configure, there’s less tolerance for spam in enterprise
- Channel growth and discovery
- More strict requirements about how bots get shared to channels
- Initiate a specific channel for your bot
Distribution is a real challenge. “A chicken and egg situation,” explains Elias. Instead, Drift focuses on the final mile. Using web site widgets, they created their own channel where every message has branding and shows support in Slack.
Another classic question in the development cycle: Are you a feature or a complete solution? “You want to be a complete solution. It’s easy to build one dialogue. You have to build a lot. We started as a human-driven messaging platform and then we started building tools,” Elias explained.
Today, they identify as a routing bot, meeting bot, and a support bot.
Kyber turns messages into to-dos, enabling knowledge workers to manage their day-to-day without ever leaving Slack.
Paolo suggests that a successful framework for designing the right type of experience begins with the understanding that different bots deliver different conversational experiences, from actionable notifications to in-line chat.
For enterprise bots, think about the different types of interaction that can be replaced with conversational bots.
- Human to human interaction (contact service provider, question/answer)
- Human to machine (system of record, input/output query) – Although this is a bit more complex, requiring a trained employee.
Chris stumbled upon the concept that eventually became Obie when he was building a document knowledge sharing solution at a previous company. Turns out, it was a great idea to a major pain point for knowledge workers, but people just don’t buy software. Obie, as a bot, allows people to access information from a place they already occupy using an interface they were familiar with: Chat.
“It’s never too early to introduce a pricing model,” Chris explains. “Having a public pricing model can offer learning opportunities. We have to stop thinking about ourselves as add-on technology, but as enterprise solutions, and price accordingly.”
CEO of Zoom.ai
Roy is trying to change the world of executive assistants. Today, 99 percent of employees don’t have an EA assisting them.
For Roy, accepting that a bot may not necessarily be profitable is part of the game for developers and creators right now.
Roy’s key lessons so far:
- Profitable Bot business doesn’t exist, and I don’t consider us a bot company, but rather an enterprise company
- We are not a chat bot company – there’s no training but there is a UI
- Trough of disillusionment will take out 90% of bots
- Profitability doesn’t exist at this stage, you’re paying to educate
- There are a lot of unknowns right now
- Focus on the enterprise – target market is employees, maybe HR
- They have scale and money
- Not throwing in a new thing – that freaks them outTraditional enterprise sales models apply