September 20 by Kingman Tang
A few months ago, I co-presented with Kevin Louie, Sales Systems Manager, Twilio, at Accelerate 2017, Apttus’ annual user conference, about Twilio’s use of virtual assistants in the enterprise. The presentation centered around how conversational interfaces, through virtual assistants, are opening opportunities to increase user adoption and productivity, and shorten process cycle times. According to Gartner (client access required), these interfaces will become a design imperative for business applications in the next 12 months. Satya Nadella, Microsoft’s CEO, famously declared at last year’s Build Conference that, “Bots are the new apps”.
I’ve been talking about the what, how and where of virtual assistants so we are now going to focus on its practical application. Kevin Louie and I discuss how his use of AI and virtual assistants are changing their business operations. This is just the beginning of their journey to the conversational enterprise.
An Interview with Kevin Louie, Twilio
Kingman Tang (KT): Thanks for doing this interview. Set the stage for us. Give us a background of Twilio and how Apttus is supporting your operations.
Kevin Louie (KL): Twilio is a cloud communications platform as a service company based in San Francisco, California. Twilio allows software developers to programmatically make and receive phone calls and send and receive text messages using its web service APIs.
Twilio has been experiencing rapid revenue and employee growth. And with a June 2016 IPO, life as a publicly traded (NYSE: TWLO) company requires us to hold ourselves to an even higher standard of operational excellence.
We have been using Apttus’ Contract Management product for our operations and we are now using Apttus’ virtual assistant, Max to further remove operational barriers as we continue growing.
KT: With the buzz machine in full hype mode, I want to get your perspective on Artificial Intelligence (AI). How did you filter out the noise from the benefits and potential of AI?
KL: What caught my attention about Max wasn’t the AI or Siri-like interaction. It was a burning pain point I was experiencing at the time. We have a very involved contract approval process and a good number of our (sales) reps were struggling with adhering to the process. To be fair, we’ve also changed the process a few times since going live. When I saw Max, I wasn’t seeing an AI bot or a piece of AI technology, I saw a potential solution to a problem. I saw Max as a proxy for me [more on why later in the interview] and she would have the same type of conversations that I would have explaining the process to a rep.
So how I filtered out the noise from the AI potential was I started with articulating the problem first, then I asked if AI can solve this problem. In my case, Max simply solves a problem that we tried solving with other traditional means that was not AI based. I saw that it could address enablement challenges that no amount of training material was addressing.
KL: There is one high level problem that I think Max can really solve for: operational process changes.
Our rapid growth makes it necessary for us to iterate and improve on our processes. However, that means sellers have difficulty staying attuned to all the process changes that are happening. I’d like to see a world where our reps are no longer in the business of memorizing our ever-changing processes and are just having a conversation with a virtual assistant about what they need help with and what needs to be done.
Within this high-level problem, there are related, smaller problems:
1. Training – hard to stay current on the latest process
2. Agility – Not able to be nimble as we are bound by the seller’s ability to adapt to a new process.
KT: And these problems are indicative of your growth and evolving processes, but that is the reality of any growing enterprise.
KL: Absolutely. Changing code on a computer is instant, but changing people’s behavior takes time. Once a change in process is announced, it takes people time to digest the change, they may have questions, and there are different interpretations of the changed process. For a fast growing organization, by the time this internalization is complete, it may soon be time for a new process.
Right now, I do take into account the number of times we change processes on our reps. And that impacts our ability to improve business operations on a rapid cadence. We try rolling out changes to our Quote-to-Cash process once a quarter instead of once a month or even once every two weeks. To overcome this limitation, we roll out incremental, “minor” changes that do not affect training – these changes come in the form of less clicks or minor reduction in process steps. Despite my efforts, I still notice reps retaining legacy processes that we have long retired (but that is human nature).
KT: Given this context, how did you make the business case for a virtual assistant like Max?
KL: I selfishly wanted Max because my team has been doing one to two hours of training per week and holding ad hoc refreshers for people to relearn the process because they missed it or they forgot. I felt a lot of time was spent on the enablement effort, but it was not sticking. And I want to get out of the training business.
We hire great reps who, right out the gate, know how to engage in complex selling cycles with large companies, but they just don’t know the Twilio process out the gate. So, I got to thinking. What if a rep didn’t need to know the Twilio process? What if we can remove the operational barrier of understanding processes? Would they be able to focus more on selling and closing the large enterprise deals that we hired them for? Would removing the training barrier enable our Quote-to-Cash team to be even faster in the way we roll out changes?
KT: Describe how you are using Max.
KL: As a background, some customers and prospects do not feel comfortable initiating a discussion with Twilio unless we have an executed Non-Disclosure Agreement (NDA) in place. With Twilio’s growth rate, the request for an NDA happens fairly often. Our reps create an NDA with the assistance of Max guiding them through fields to fill out. If the process requirement evolves, we would only need to adjust Max to ask only what needs to be filled in. At the end of the flow, the completed NDA is in a PDF and sent for e-signature. This is completely self-served by the sellers and our legal team is only involved for exceptions or redlines.
KT: What kind of benefits are you expecting to see with Max?
KL: For our NDA use case, I’d like to see every single one of our reps self-serving NDAs. On a broader level though, I am hoping to dramatically reduce the amount of time it takes reps to do error-free contracts. This should also free up Legal, Finance, and Sales Ops to do less tactical work and more strategic work.
KT: Going forward, what are other things you want to do with AI and Max?
KL: NDA is just the start. I want more self-service order forms, which will free up more time for our operations folks. I’d also like to explore fast quoting with Max – with a few questions over a Slack (Slack is Twilio’s collaboration / communication tool) conversation, it’d be great to create a quote for both the rep and the customer to use.
Using Max, search for the appropriate Twilio point of contact for any given customer based upon what you are trying to accomplish. For example, if someone wants to know who the “business point of contact for customer ABC is”, I see Max asking clarifying questions, understand what “business” means so that they can get to the right Twilio point of contact.
Max taps into machine learning algorithms to advise and recommend to sellers deals that they should not commit to or surfacing deals that reps don’t know about. These recommendations are based upon similar customers and or similar verticals.
KT: What advice would you give to people thinking about AI and virtual assistants? How can they get started?
KL: Crawl, walk, run. Start with simple use cases (low level of effort), but high impact (increase value). That is why we started with the NDA creation process.
If you are thinking about AI or a virtual assistant, I suggest starting with a meaningful problem you really want solved. Have I tried solving it with conventional approaches? If yes, then can AI or a virtual assistant help?
At the end of the day, AI is a means, not an end. In our case, we see Max as a way to solve the training and process problem that we have not been able to solve yet.