James Evans is the Founder & CEO of CommandBar, an AI-Powered user assistance platform that is designed to assist product, marketing, and customer teams. They offer offer an AI co-pilot that only answers questions based on your content. And when it doesn't know something it defaults to an experience you control – like starting a support chat.
What initially attracted you to computer science?
I was originally a math major. Not for any particular reason, but because I thought math was the hardest major. Turns out math is really hard at Princeton, and I was much better at computer science than I was at abstract algebra.
What kept me in computer science was that it felt like a really unique field, where you could dive deep into theory (the subject of my independent work) but also learn how to build products.
In 2019, you launched codePost, a feedback tool for teaching computer science and programming. What inspired this company and what did you learn from the experience?
codePost was a kind of “practice startup” for me and my co-founders. We knew we wanted to start a company together, and codePost was a project we started in college that already had users (including Princeton) who were using it to grade computer science coursework.
We worked on it for a year before shifting our focus to other endeavors– it still exists! https://codepost.io. We learned with codePost that shipping fast and being super responsive to user questions and feedback was a competitive advantage. We also learned that when you do that, your product can get complicated and hard to use quickly. That insight is actually what led us to build the version of Commandbar, which was a widget we embedded inside of codePost to make our product easier to learn and use.
In your view, what are some of the common issues with the current landscape of digital tools?
Software is becoming increasingly complex, and companies use many different tools, making it hard for users. The expectations that end-users have are very high due to user-friendly consumer apps, and nobody wants to read lengthy manuals for everyday software.
At the same time, the way software looks and feels hasn’t changed. And the main way to help users is…pop-ups. Those little messages that pop up and interrupt you when you’re trying to get things done. Sometimes they’re helpful, but most of the time users try to close them as fast as possible.
Can you share your vision for a new generation of user-centric tools?
We started CommandBar because the current state of human-computer interaction left room for improvement. Through conversations with users, we repeatedly heard that pop-ups felt annoying – the opposite of helpful!
Our approach centers on providing users with a significant enhancement in their software tool experience by honing in on user intent – what they aim to achieve with a software tool. Detecting intent can take an implicit form, such as deciphering user confusion from their in-product behavior, or an explicit one, like responding to user queries and searches.
The most effective user-centric tools aim to personalize the user experience by comprehending user intent and refraining from inundating the user with company-centric information that they may not find relevant. This means fewer generic pop-ups and more personalized guidance to enhance the overall user experience.
How did this vision transform into launching CommandBar?
CommandBar is our attempt to be a layer on top of all products that can help users by detecting and acting on their intent, using AI. For example, one of our products is an embedded user assistant that can answer user questions and carry out tasks for them, meaning they can get things done without needing to learn how a product works. Another of our products “nudges” users towards paths, features, or walkthrough content that seems relevant to their intent.
It started as a single in-product widget: a search bar that lets users search for anything inside a product, kind of like Apple’s Spotlight. From there, we branched out to other user assistance widgets.
One of the core CommandBar products is Copilot, a tool that allows users to easily integrate a chatbot on a website. Could you share some details on how this works in the backend?
What we first launched as HelpHub AI is now called Copilot. You can think of it as a “Copilot-as-a-service” product that lets any software product – web app, mobile app, website, or desktop app – train and deploy an embedded user assistant.
Copilot begins by ingesting any existing content that is designed to help users – documentation, blogs, guides, walkthrough videos, and API documentation.
At this point, Copilot undergoes a training process and is prepared to assist users in two key ways:
- Answering user questions.
- Performing actions on the user’s behalf.
Over time, Copilot learns about user preferences by analyzing their chat history and in-product interactions. For instance, it can identify that some users prefer step-by-step tutorials for understanding actions, while others may favor concise, text-based answers. Some users might even prefer completing actions directly within Copilot.
In addition to user preferences, the goals of the company using Copilot are also considered. Our clients, typically software companies, deploy Copilot to assist their users. Sometimes the objective is broad, focusing on overall user assistance, while other times, companies aim for specific outcomes. These goals can range from reducing the volume of support tickets users submit to their human support agents to encouraging more users to upgrade to a paid version of the product. Copilot integrates these goals into its assistance strategy, benefitting both users and the company serving them.
Through the first few weeks of Copilot, we have already powered hundreds of thousands of end-user chats.
What are some additional products that you plan on launching?
We just officially launched CommandBar Mobile Copilot! This is a huge release for us because it brings all of our assistance experiences to way more users.
Most companies that we work with have a big mobile presence—and this assistance is vital because the screens are smaller, there’s no keypad or mouse, and often users are on the go and have less time to spend understanding an interface.
Up next, we’ll be launching modules designed specifically for Sales and Customer Success teams to leverage CommandBar to design customized assistance flows for key prospects and customers.
What do you think will be the next big breakthrough in AI?
I’m biased but I’m most excited about models, like Adept’s Fuyu, that will become general-purpose software manipulating agents. There are a lot of applications of this beyond the scope of CommandBar, including training RPA-style bots to handle tedious tasks for users In the user assistance space, I think these models are going to allow tools like our Copilot to perform more actions on behalf of users and do a better job of walking them through how to perform actions on their own. Like the second coming of Microsoft’s Clippy!
In a world that is exponentially changing faster, how can businesses best remain competitive and even outperform the competition?
It’s been super cool to see big companies like Microsoft react so quickly to the AI wave. It’s exciting to move fast again! At CommandBar, we've encountered numerous Fortune 500 companies seeking guidance on crafting AI strategies for their products.
I believe this marks a significant turning point in corporate governance. Large companies have realized they can move quickly and effectively when there's alignment throughout the organization.
To maintain and build upon this momentum, the key, in my opinion, lies in the continuous evolution of your AI strategy and beyond. Don’t launch a few AI features and call it a day. Instead, businesses must stay at the forefront of emerging developments and consistently roll out experimental initiatives. Treating this moment as a one-time event rather than a new standard for product adaptability would be a significant misstep.
Thank you for the great interview, readers who wish to learn more should visit CommandBar.
- The Black Box Problem in LLMs: Challenges and Emerging Solutions
- Alex Ratner, CEO & Co-Founder of Snorkel AI – Interview Series
- Circleboom Review: The Best AI-Powered Social Media Tool?
- Stable Video Diffusion: Latent Video Diffusion Models to Large Datasets
- Donny White, CEO & Co-Founder of Satisfi Labs – Interview Series