Interview

Interview with AI Bot Conversation API PubNub’s CTO Stephen Blum

Stephen shares his insights on PubNub’s Real-time Communication Platform, and how AI bot conversation APIs can improve cost efficiency, CX for businesses

1. Tell us how you came to be the co-founder and CTO at PubNub? What does your role entail?

As one of the founders of PubNub, I drove the initial architecture Building a core engine that has successfully scaled some of the largest real-time deployments in the world is still my main focus as CTO. The CTO role is different at each company, as many offer technical oversight and guidance. At PubNub, our architecture team is strong. We think of it as a counsel of CTOs operating a global network delivering trillions of JSON payloads to 780M mobile devices.

2. Can you tell us about how you made your career in software development?

I’ve been a software entrepreneur my whole life, starting my first company LTSun at age 17, and then joining Microsoft after its acquisition. I was also the chief architect for TrustedOpinion.com, a 50-language social recommendations site that grew to over 1 million users across over 100 countries. My co-founder Todd Greene and I launched PubNub in August 2010 to let companies quickly create sophisticated multi-user mobile and web apps that scale globally.

We are in the API market, and our offering allows developers to deliver custom JSON messages to mobile and stationary devices within thirty milliseconds. This enables critical apps like tele-health services to work successfully connecting doctors and patients instantly. We also see gaming companies using PubNub for multiplayer connectivity, and saw those opportunities early on. We did not know which market segments would be bigger for us, but of June 2020, healthcare telemedicine is the biggest growth segment – no surprise as this market has experienced huge growth globally this year during the pandemic.

3. How do you think sentiment analysis technology is upgrading the Communications sector?

Machine learning is shaping the way we communicate, and the automatic detection of sentiment allows us to course-correct our communications. This is a little scary. The keyholders to our future of verbal and textual communication is in the hands of the AI developers.

4. Why do you think businesses should make a move to AI-driven chatbots from their traditional way of conversing?

AI-driven chatbots create better margins by helping businesses automate, improving cost efficiency. Chatbots are now nearly as effective as a human behind a keyboard, and they cost almost nothing, they never sleep and can support unlimited inbound inquiries. Every business should invest here, especially for customer service and AI assistants.

Chatbots work best when they appear to operate at an artificially slower speed, roughly the same speed as a human operator. While there are surely circumstances where they need to be decelerated (a 100-word response to a complicated question appearing in half a second is disconcerting), the reality is that the sophistication required to get the interaction right demands a level of system speed and direct response that can be tricky to deliver on a traditional infrastructure. The AI must learn to be human.

5. How do you think the use of Natural Language Processing has played a vital role in developing a digital infrastructure for bidirectional communication networks?

Human input bandwidth is high, and we can readily consume information from various sources such as videos and articles. However, it takes longer to create that information. With communication between users online, the same problem exists. Forming words to best match ideas is imprecise. Once we add NLP, this information sharing can be made compact and made better, faster, and can accelerate our human output bandwidth.  This improves the rate at which we communicate, so businesses can operate at higher efficiency.

NLP can analyze massive amounts of data inputs and derive insights from them, whether it be responding accurately, finding patterns, determining the language, or gauging the sentiment of it.

A common example for NLP would be to analyze a document like ‘The Declaration of Independence’ and return findings on it like an estimated period of writing, the language it has been written in, the tone, the complexity of the writing, etc. Will we find hidden loopholes in our constitution using AI discovery techniques?

6. Can you tell us about the ways that PubNub takes care of security and protection of the solutions against any cyber threats?

PubNub has security measures at the network, message, channel, user, and key level, spanning multiple categories. The most important security feature is end-to-end encryption. Cyphers and private keys are not known to PubNub, and data is encrypted on-device before transmission to the network.

Security must be looked at from a different angle when applied to data streams. Our continual infrastructure patching, auditing, and encryption ensures that the network is secure, but what if one of your data stream endpoints, like a smartphone, becomes compromised? PubNub lets customers block any device as soon as a security threat or unauthorized access is detected. Furthermore, traffic can be separated into channels with different access levels, restricting channels requiring the highest levels of clearance to a small subset of devices, lowering the attack surface and reducing exposure.

7. What advice would you like to give upcoming startups in this sector?

There are a lot of opportunities, and business opportunities are everywhere. There’s a trick though. Spend 90% of your time on promotion, and the rest of your time on product development with one key feature.

8. What digital innovation in the tech space do you think will leave a mark in 2020?

Certainly, AI’s de-anonymization capability has made a mark. Your words, your actions are all classified and identifiable with AI.

9. How do you keep up with the rapidly developing tech world?

We’ve been the leader in data delivery since early on, and our key difference is the globally distributed messaging network we offer. Mobile users connect to the closest edge, which improves performance and user experience. Today, our technology enables faster communication that won’t be commonplace for decades. Thousands of companies use PubNub’s Real-time Communication Platform and APIs as the foundation for online chat, live events, geolocation, remote control, and live updates, on a massive global scale.

10. Can you tell us about your team and how it supports you?

We are boilerplate. We have an amazing product team that shapes and manages customer feedback and sales opportunities to create new and advanced features. The PubNub Engineering team makes it work at a global scale approaching one billion mobile devices. Collaboration is critical, as the teams are all remote, but we’ve been able to maintain our high output of projects with online teamwork and maintaining the lines of communication.

11. What book are you currently reading?

Recently, I have been keeping up with the latest technology. Books are typically one or two years behind.

The rate of innovation outpaces books, especially in technology. Foundational books are great references and remain relevant. The future isn’t written.

12. We’d love to get a peek into PubNub’s chic and fabulous work culture! Can you share some pictures of get-togethers and office fun activities?

While we moved to a new office earlier this year, we all work online now and over video conferencing. Because PubNub creates remote enablement, our team experiences new ways we can collaborate online. Often to our own amusement! It’s been interesting to see how our company culture has both evolved and migrated without a common meeting place on a daily basis.

Stephen Blum, co-founder and CTO at PubNub, showed his technical chops as a software entrepreneur at 17, when he started his first company LTSun, later acquired by Microsoft. He was the chief architect for TrustedOpinion.com, a 50-language social recommendations site that grew to over 1 million users across 100 countries.

Related posts

Interview with Jim McGowan, Head of Product of AI-based BI Tool ElectrifAi

AI TechPark

Interview with Caroline Hynes, Director of Product for Zendesk AI-Driven Customer Engagement Platform

AI TechPark

Interview with Roger Beadle, Co-founder and CEO at Limitless

AI TechPark