20VC Building An AI Company For The Long Term How Humans & Machine Learning Can Work Together What Applications Are The Conversational Interface Best Suited To with Maran Nelson, Founder & CEO @ Clara Labs

In a conversation with Harry Stebbings on "20 Minutes VC," Marin Nelson, co-founder and CEO of Clara Labs, discusses the evolution and future of digital assistants. Clara Labs, known for its conversational AI that intelligently schedules meetings, has garnered investment from figures like Kent Goldman of Upside Partnership and Greg Brockman of OpenAI. Nelson, also the founder of Interact, shares her journey from a humanities background to tech entrepreneurship, emphasizing the importance of customer-centric development and robust machine learning environments. She highlights the challenges of integrating AI into user-friendly products and the incremental nature of technological advancement, with a focus on conversational interfaces as the most natural progression for complex tasks. Stebbings and Nelson also touch on the role of human feedback in AI training and the anticipated integrations of Clara with other business systems to enhance productivity and responsiveness.

• Harry Stebbings introduces the special Founders Friday episode of the 20 minutes VC.
• Harry mentions his activities on Snapchat and his writings on Mojitovc.com.
• Marin Nelson, cofounder and CEO at Clara Labs, is welcomed to the show.
• Clara Labs is highlighted for its conversational intelligence for scheduling meetings.
• Marin is also recognized for founding Interact, a community for young makers and technologists.
• Appreciation is expressed for the support and respect for Marin from the community.
• Harry gives a shoutout to Kent Goldman and Greg Brockman for their roles in Clara Labs' success.
• The podcast briefly promotes Simba hybrid mattresses and Sirius Insight, a productivity tool for sales professionals.

With me, Harry Stebings found posting mojito masterclasses on Snapchat at H stebbings and writing some hopefully more pensive and academic thoughts on Mojitovc.com.

This quote introduces Harry Stebbings and his activities outside the podcast, indicating his engagement with his audience through different platforms.

Marin is the cofounder and CEO at Clara Labs, the truly human interface that uses conversational intelligence trained on high quality data to schedule your meetings.

This quote introduces Marin Nelson and her company, Clara Labs, highlighting the company's focus on conversational intelligence for scheduling.

• Marin is originally from Texas with a background in humanities, psychology, and neuroscience.
• She conducted undergraduate research on human intelligence.
• Her cofounder and best friend since age 15 was an early iPhone app developer and studied robotics and machine intelligence.
• The cofounders had ongoing discussions about the future and what they wanted to create, leading to the conception of Clara.
• Clara was founded on the thesis of creating accessible and scalable human-like intelligence.
• The inflection point for Clara was the realization of the widespread issue of scheduling.

I always saw myself as somebody rooted squarely in the humanities, loved people, studied psychology and neuroscience, and did my undergraduate research on intelligence, human intelligence.

Marin discusses her academic background and interests, which laid the foundation for her human-centric approach to technology.

Shortly after school, constant conversations about what we expected to exist in the future that didn't exist today, how we could contribute to bringing it about, and had several conversations that led and fermented into our conception of Clara.

This quote explains the ideation process behind Clara Labs, emphasizing the founders' vision for the future and their desire to solve problems through innovation.

• Marin addresses the tendency to anthropomorphize technology and the historical context of digital assistants.
• She references the personal digital assistant (PDA) from the past and compares it to modern smart devices.
• The challenge of explaining the role of Clara as more than just an assistant is acknowledged.
• The conversation touches on the incremental evolution of technology and user interfaces.
• Marin cites Snapchat and Pokemon Go as examples of technologies that incrementally introduced new behaviors to users.
• She suggests that Clara Labs is part of a similar incremental innovation, enhancing existing software suites with intelligence and agency.

We have this concept of a digital assistant right now, and we think that it's new and that we've kind of recently come upon it.

Marin provides context for the digital assistant as a concept, noting that it is not a new idea but has evolved over time.

What we're doing is just making this core software suite that you spend your life in, that you spend hours a day in, much, much smarter than it has been before by growing to understand you, by learning your preferences, by being able to have the agency to take some actions on your behalf.

This quote clarifies Clara's role as an intelligent enhancement of software people regularly use, aiming to make daily tasks more efficient through personalized understanding and action.

• Marin believes that technological changes are more incremental than radical.
• She emphasizes the importance of familiarity in user behavior patterns for the success of new technologies.
• Incremental innovations are seen as wise strategies for introducing advanced technologies to the public.
• Marin highlights the significance of product focus in technology companies like Snapchat.
• The discussion implies that Clara Labs is taking an incremental approach to introducing conversational intelligence to users.

I think everything tends to be much more incremental than we like to think of it before it happens.

Marin discusses the gradual nature of technological evolution, suggesting that significant changes often occur through a series of small, incremental steps.

Snapchat has this relentless product focus, and now, of course, they're the people introducing the idea of augmented reality to tons and tons of people.

Marin uses Snapchat as an example of how a company can successfully introduce new technologies by maintaining a strong focus on product development and user experience.

• Conversational interfaces align closely with existing user behaviors, making them an elegant solution.
• They avoid the uncanny and unfamiliar, providing a seamless transition from current habits.
• This approach is based on the premise that users are already accustomed to conversational interactions.

"It's elegant in that it is not uncanny, challenging, unfamiliar. It's really just a step further from the behavior that you don't think twice about."

The quote emphasizes that conversational interfaces are a natural extension of behaviors that users are already comfortable with, which reduces the friction of adopting new technology.

• Scheduling was identified as a tedious, frustrating, and error-prone task, leading to the development of Clara.
• The challenge of creating a simple and accessible interface for human negotiation led to the insight that conversational interfaces may be superior for complex tasks.
• Traditional software apps have not been successful in solving the scheduling problem due to the complexity involved.

"We started trying to build a normal app, a software app of some kind, to solve the scheduling problem...and found that it was really challenging to think about designing an interface that was going to make this negotiation, this human negotiation, simple and accessible."

The quote explains the initial attempts to create a traditional app for scheduling and the realization that a conversational interface might better handle the complexities of scheduling negotiations.

• Conversational interfaces are most useful when the complexity of a task makes traditional graphical user interfaces less effective.
• Current bots lack sophistication to handle nuance or complex tasks, limiting their application to simpler tasks.
• The tipping point for preferring conversational interfaces over graphical ones is when the task complexity warrants it.

"Language interfaces are really important if there is a sufficient amount of complexity such that language is the best interface to resolve that complexity."

This quote highlights the importance of language interfaces for complex tasks where traditional interfaces may fall short.

• The tipping point involves the user's ability to perform tasks without excessive information input.
• Users prefer fast and intuitive interactions, with varying willingness to invest time upfront for ease of use later.
• The preference for interface types can vary based on user habits, such as those who use keyboard shortcuts.

"I think it has something to do with your ability to know enough information about the person such that you can act on their behalf without having to get too much information from them."

The quote suggests that conversational interfaces become preferable when they can efficiently act on behalf of the user without needing too much input, streamlining the user experience.

• The current categorization of AI as a single category is frustrating and misleading.
• AI and conversational interfaces should not be viewed homogeneously, as they vary greatly in their applications and complexities.
• It's important to develop a better vocabulary to distinguish between different types of AI and conversational tools.

"Most of these things in this category are so painful. Most of these assistant things or whatever are hyper tedious."

This quote reflects the speaker's frustration with the current perception of AI and the need for better differentiation between various AI applications.

• User experience is paramount in building any company, including AI-focused ones.
• A company must understand and meet customer needs, which may involve manual processes initially to learn and scale the desired user experience.
• A solid environment for machine learning is crucial for a long-term AI company.

"The most important thing is the user experience, period. Right? As with everything else in the world, if you're trying to build a company, you have to put your customer at the absolute heart of everything you do and work backwards from what it is that the customer needs from you."

The quote underscores the importance of centering the customer in the development of a company's product or service, especially in the context of AI where user experience is critical.

• Young companies are integrating AI from the start, creating environments conducive to machine learning.
• Legacy companies like Zendesk may face challenges updating their systems to support AI as effectively.

"Vantage to many of the younger companies that have started from their foundation with AI at the center of everything that it is that they're doing, because they have created much better environments to support machine learning there than, for example, like a Zendesk or something that may have started a long time ago."

The quote highlights the advantage newer companies have by incorporating AI from the beginning, suggesting they are better suited for machine learning compared to older companies.

• A good machine learning environment requires effective feedback loops.
• The ability to recognize correct and incorrect predictions is crucial for intelligent training.
• Siri's recent feature allowing users to report errors is an example of improving feedback loops.

"It means good feedback loops. It means you have to understand when you predicted it right or wrong, with enough reliability that you can train intelligently..."

This quote explains that a good machine learning environment is characterized by the presence of strong feedback mechanisms that allow for accurate assessment and training of AI models.

• Siri added a feature for users to report errors, enhancing its feedback loop.
• Clara, an AI system, is designed with strong feedback loops for learning.

"They added something recently to the bottom of the screen such that you can kind of report the error, though you don't have much of an incentive as a single user to spend additional time to report Siri's error to the Apple team."

This quote describes a practical example of a feedback loop in AI, where users can report errors to improve the system, despite the lack of individual incentive to do so.

• Clara's AI platform processes emails and takes actions based on confidence levels.
• Uncertain predictions are reviewed by contractors, called Clara remote assistants, who are well-acquainted with the software.
• The contractors provide complex feedback, not just simple labeling tasks.

"The way that Clara was always kind of envisioned to work, and the thing that we are really proud to have achieved over the course of the past two years of investment has been building a platform wherein a message comes in an email of some kind..."

The quote explains Clara's operational model, emphasizing the integration of AI with human oversight to ensure accurate processing and response to emails.

• Humans will continue to be a part of AI systems until strong AI eliminates the need for human tasks.
• Google's approach to AI learning involves massive data and simple labeling tasks by contractors, which can result in noisy labels.
• Clara's contractors provide valuable, complex feedback due to their depth of context and understanding of the system.

"I think they will always be involved. We get asked that frequently, and it makes perfect sense for us, though, until we have some kind of strong AI that completely obfuscates the need for human work."

This quote underscores the current necessity of human involvement in AI systems and suggests that it will persist until AI technology advances sufficiently to replace human tasks.

• Marin Nelson's favorite book is "East of Eden" by John Steinbeck for its philosophical undertones disguised as fiction.
• Scheduling is a natural language task well-suited for machine learning due to its repetitive nature and low variance in responses.
• The biggest takeaway from Y Combinator for Marin Nelson is the nurturing and trust-building aspect for early-stage companies.

"My instinct was east of Eden by John Steinbeck... It's really deliberate about life and what it means to live well."

The quote reveals Marin Nelson's preference for literature that intertwines philosophical concepts with storytelling, particularly noting "East of Eden" for its exploration of life and morality.

• Machine learning is ideal for tasks with repetitive patterns and constrained variables.
• Scheduling involves a limited set of variables, allowing for more effective machine learning applications.

"So scheduling, luckily, is very much one of them. Repetitive. Right. You want the variance in response to be low, ideally."

This quote identifies scheduling as a prime candidate for machine learning due to its repetitive nature and predictable variables, which facilitate the training and accuracy of AI models.

• Y Combinator builds a strong relationship with its companies by being involved at an early stage.
• The trust established through this relationship is considered invaluable by founders.

"Yeah. My impulse is, like, they just really do want to take care of their companies."

The quote reflects Marin Nelson's positive experience with Y Combinator, emphasizing the supportive and trust-based relationship that the accelerator fosters with its startups.

• Machine learning and human learning are similar in that repetitiveness and exposure to similar problems enhance intuition and problem-solving abilities.
• Y Combinator provides an environment where this type of learning is facilitated through exposure to numerous companies at similar stages.
• This experience allows for the cultivation of good counsel based on familiarity with common challenges.

"The same things that machines are best suited to learn, humans are largely best suited to learn, which is the more you can interact with the same types of problems, the better an intuition you'll get for how to solve the problems."

The quote emphasizes the parallel between machine learning and human learning processes, highlighting the importance of repetitive interaction with problems to develop problem-solving intuition.

• Marin Nelson's preferred reading materials include Aeon and Natilus, which cater to her theoretical interests.
• She distinguishes herself from other guests by emphasizing her habit of reading books.

"Aeon and Natilus. Okay. Are really both. Interesting."

This quote reveals Marin Nelson's sources of intellectual stimulation and suggests these publications are valuable for those interested in theoretical and in-depth discussions.

• The main challenge in ML is creating a user experience that is reliable and meets high standards of accuracy.
• Achieving high reliability in predictions (95%+ accuracy) is necessary for a good user experience, but it is difficult to reach this level.
• The "catch 22" in ML is that good user experience depends on reliable features, which in turn require accurate machine learning predictions.

"To have a good user experience, you have to be very confident that the feature that you're shipping is going to work right."

This quote underlines the importance of reliability in user-facing features and the challenge of achieving high confidence in machine learning predictions to ensure a positive user experience.

• 2016 was a significant year for Clara with the launch of Exo, a machine learning platform.
• Exo serves as a suite of productivity tools, including email, maps, and calendar clients, and it learns from user feedback to make predictions.
• Clara's goals for 2017 include integrations with other systems, such as booking conference rooms and applicant tracking systems, to enhance its service offerings.

"So today, Clara doesn't book conference rooms, for example, which we are pretty eminently going to be doing and obviously keeps us out of a lot of larger companies."

The quote outlines a specific goal for Clara in 2017, which is to expand its functionalities to include booking conference rooms, thereby increasing its appeal to larger companies.

• Harry Stebbings expresses gratitude to Marin Nelson for her insights and to Siobhan at Bloomberg Beta for making the interview possible.
• Harry shares his personal goals for the new year, including better sleep and increased productivity, and endorses products that align with these goals.

"I watched your incredible growth in 2016 and I was told that it'd be fantastic to have you on the show, but you've absolutely blown away all expectation."

Harry's quote conveys strong appreciation for Marin Nelson's contributions to the show and acknowledges her company's impressive growth in the previous year.