Welcome to our newest season of HumAin podcast in 2021. HumAIn is your first look at the startups and industry titans that are leading and disrupting ML and AI data science, developer tools and technical education. I am your host, David Yakobovitch, and this is HumAIn. If you liked this episode, remember to subscribe and leave a review. Now into our show.
Today’s guest on HumAIn is focusing on how to revolutionize the way that we consume video, as both, consumers and enterprise companies. Today’s guest is Humphrey Chen, who’s the CEO and co-founder of CLIPr.ai, joining us from Seattle. Humphrey, thanks for joining us on the show.
Thanks very much for having me.
Well, I love the whole space of consuming media and changing how we consume media. Throughout the last couple of years of being digital, I look at my screen 16 plus hours a day. There’s so much video, there’s so much content, and sometimes I just want to skip through content. I want to turn off my video. I want to consume things quicker. So I think a lot of these pain points I’ve experienced your team’s working on. Can you tell us a little bit about yourself and what is CLIPr.ai?
Absolutely. So what we’re building leverages stuff that I was working on when I was at Amazon and that was in the computer vision team. And in AWS, it was my job to help developers to see and hear at scale. So flash forward to the pandemic, basically when people started to get more and more personal videos, this is like work-related school, related events, related people, were starting to get backlogs of content and just not enough hours in the day to actually get caught up on that content.
So, for CLIPr our operating premise is that not all minutes of video are created equally nor are they equally relevant to everyone. Yet, when you hit the play button you’re at the mercy of everything behind that play button. So what CLIPr does is we use machine learning to fully index that video, whether it’s Zoom, Meet, Teams or Events, and we make it fully searchable. For premium content, in this case, typically events content, we will actually generate an automated table of contents. So if only 5% of that session is actually relevant to you, you can actually just watch that 5% and skip the rest. If there’s another 5% that you actually might need to know, then you can actually search and find the other 5%. If we can actually save 90% of the time required to actually get caught up, mission accomplished. So, our goal in CLIPr is to help you to do more and watch less. So, before COVID the percent of meetings and things that were actually getting recorded was like literally fractional single digit percent.
It’d be like all hands and like really high value things or events. Now we’ve reached a point where it’s like double digit percentages of meetings and events getting recorded. All the events are getting recorded, but, basically, the tools to get caught up on things just have not caught up with the fact that there’s so much video now. This video is actually very different from the video that we watch on Binge, on Netflix, or FireTV, or all these other things there. Every minute of video was part of the story and it’s like finally curated every day video is not. That’s where it’s not as efficient and CLIPr’s goal is to make it as efficient as possible so that you can just see what you need and want and get done what you need as well.
I can see so many strong cases for the technology at CLIPr.ai. I’ve worked in the last few startups with client technical services, through training and enablement with enterprise companies. When I do a 60 minute Webinar I’ll cover topics such as column stores, row stores, sharding, reference tables, indexes. I can imagine that a participant may not want to watch all 60 minutes of the video. They may want to watch, just reference tables or specific sections. It’s so frustrating. I even think back to all the recordings I watched today. It’s like, skip ahead, 10 seconds, 15 seconds until I get to the point. It’s really inefficient. So, the technology that your team’s building is speeding up and accelerating more efficient automations and that sounds like it’s going to be helpful for both consumers and enterprises.
Absolutely. That inefficiency right now, when you’re creating content in an event you’re trying to like to cater to a wide audience. So it ends up buying by design, having to say many different things, but the audience usually only ends up having specific things they care about as opposed to all those things, that also applies in everyday meetings. So one of the things that we’ve been kind of driving right now is this concept of real-time optional. So this idea where when people get invited into a meeting, there’s going to be a required attendee and there’s an optional attending. But this new category that we’re thinking about is real time optional and your real-time optional.
If you’re not actually in the meeting and changing the trajectory or the decisions that are being made in the meeting and you by design are a fly on the wall listening for things that you may or may not need to know.
In that situation, that’s a perfect opportunity for CLIPr to basically process that meeting and then you can just get quickly caught up on that, like real-time optional meetings. Because if you’re a fly on the wall, maybe there’s actually something else that you should actually be focused on and getting done, instead of actually sitting around waiting for that phrase or that sentence that ends up being relevant to you.
So, that’s actually another way that we’re trying to solve for video fatigue, attending just the meetings that you want. Right now, when people record things, and they send you the link, It’s not like the first thing that you actually want to watch is probably like the last thing you want to watch, because to your point earlier, it’s not easy to get caught up.
So the tools that we were providing ended up making it a lot easier to get caught up. So if you can efficiently catch up, and actually, if the person who sent you that content can also see that you, actually, did get caught up and see exactly which parts of the video you saw, that also ends up being really informative, because at that point there’s no leap of faith of like: Oh, did that person actually watch, or did that person not watch? And it’s like not. There’s a lot more detail involved in just clicking it and starting to watch it versus, actually, knowing which part of the video the person, actually, spent more time in and which part they actually didn’t even look at all.
So, those types of tools currently don’t exist. So for CLIPr, we’re really focused on this video insight because most people are flying blind when it comes to engagement and interaction around the video itself. So for us, it’s kind of like this category that we’ve created, this called video analysis and management.
So what we want to do is let any developer let any end-user tap into our solution to process their video they have and there’s a backlog of content that they care about upcoming, valuable content, medium valuable content. These things have referential utility, but if you can’t easily access it, might as well not be there.
So for CLIPr, we really want you to have the tools to basically efficiently find what matters, book market, share it, react to it, comment on it. So the other thing that we’re really thinking about and building here is, if you think about Google Docs or Microsoft Word and collaborating around a text-based document, those tools are very, very mature.
If you think about collaborating around a video, people immediately think about Adobe Premiere Pro. The tools and the skills required to edit a video are completely opposite of the skills and tools required for editing inside of a document. And what CLIPr’s doing is bridging the two. So, we’re actually, effectively, building a video based document type.
So now you can imagine that video recording ends up being a memorialization of that session, from which tasks can be assigned. People can be notified about different things and things can get done from it. So that’s really valuable for a high value meeting. But it’s also valuable for a medium value meeting, but it can actually be valuable for all meetings.
So this vision interacting around this like document type, and that’s also kind of like why you can also think of CLIPr as like CLIPr notes. It’s like, we want to like make things efficient and summarized, but we also want you to interact off of it. So, we’ve been putting a lot of thought into this, and it’s awesome that there is so much video now.
So, what is frustrating and we’ve heard from a lot of people is that this video and the potential inside the video is all trapped. So CLIPr is actually helping to unleash the value and the potential from that video, because it’s actually all there and machine learning actually helps us to expose that. So for us, it’s really been awesome over the past year. Just building this, piloting this with customers in all different segments. Learning about use cases, and I can get into like a couple of use cases that have really stuck that are transformative, because people aren’t really aware of what’s possible. So right now we’re spending most of our time identifying design wins that end up lighting up verticals and use cases by industry, by role, by geography.
That’s right. When we think of different industries that have been disrupted around text, around audio. There’s been a lot of changes since the 1970s, where there’s been success with live transcripts and unique audio filtering technology, but there has not been as much disruption around video. We start to see collaborations, of course, in design, where Sketch went to Figma online with collaborative design, and we’ve seen what some video editing, like the frame.io and other platforms on that collaboration. But really when we come back to videos and consuming them, the insights are still very 1970s.
You have a broadcaster, you have different screens, you’re piecing it together. It’s a very manual process. You have the different encodings, but not much of this is automated. I can think myself so many use cases. I’m thinking of the future of using Coursera. I should be able to go anywhere in the video although different segments or watch Saturday Night Live on NBC and actually go to the weekend update very quickly instead of listening to a band that I don’t prefer. So there’s a lot of opportunity, for both consumers and enterprise. You mentioned more about these use cases, Humphrey, whether some of the exciting ones that you’re working towards today.
So one of the ones that I really like is this video voice of the customer, and we’ve all had this happen. Where if you ever talk to a sales person, you’ll never hear from a sales person that the sales call didn’t go well, like they all went really well, but they all don’t end up closing. So what this particular customer wants to use CLIPr for is they’re recording all their customer meetings; and within these customer meetings, when customers provide feedback on a product good or bad, what ends up happening inevitably is when the field comes back to headquarters and provides that positive or negative feedback.
Oftentimes the messenger is accused of being sugarcoating or exaggerating, with this video voice of the customer powered by CLIPr, We end up effectively being that source of truth because that field person can actually play the exact moment of what the customer said, and we can actually analyze how that customer said it.
So that means that there is no longer any risk of the messenger conveying things inaccurately, but it also ends up being something that can actually be integrated in with a CRM record. So that later on, when the improvements to the product are actually delivered. The field person can actually later on go Mrs.Customer, Mr. Customer, you asked for this, play and Boom. Here it is. The other thing that we can do is provide a score over time around the actual probability of closing that sale based on the relative perception for the customer reaction. So that’s not a first thing, that’s a later on thing, when you actually start to get deeper into specific use cases, you actually can improve on the technology in a very specific value added way.
So, Individual use cases have been using CLIPr in a horizontally applicable way with referential utility and searching. But then when we’ve been looking for specific design wins there’s opportunities for us to layer in additional tech to actually make that use case even more valuable, so we get roadmaps that are actually specific to that use case that we then can actually work towards making happen. So that’s one use case, and another use case our first paying customer ended up being a leading ear teaching hospital in New England. And they’re using CLIPr because surgeons are really craftspeople that are focused on healing and fixing people.
And the only way that surgeons can actually improve their skills is to actually look at other surgeons who actually do the same procedure or have done it even better. You can only learn by watching. So, pre CLIPr, you would have actually been skipping and skimming plus 10, plus 10, minus 10, plus 10, then maybe searching for an audio transcript.
But if you look at it with CLIPr you can jump literally right to that moment, see that exact moment, tag that moment, categorize that moment and refer to that moment and share it with other colleagues in the hospital. So like, you can imagine for like a surgeon, high value, highly visual, high-impact not enough hours in the day. So now in that case, 21 hours of a virtual bootcamp that basically was served in 30 countries, like 270 different surgeons around the world. This hospital made their content available through CLIPr. It actually means that the whole conference can actually be referred to almost forever, because it’s actually useful. It’s not just the link sitting in an email box that they may every may go to. They, actually, can refer to it and others can actually benefit from it. So in a way, the shelf life for that content gets extended and more people can benefit from it.
There’s actually a network effect because when you are an attendee at an event, you get exposed to that content and you can then share it with others and other people can actually benefit from it, and then they can invite others to that same team and it can actually grow. And we view that as like a flavor of landing and expanding because anybody who ends up experiencing CLIPr process content ends up going: Well, hey, this is cool. I can actually use this for my own content. So that ends up being another dynamic.
When you think about the video voice of the customer, as you brought up, I find it so fascinating to think about the sales playbooks with CRMs. It’s the classic problem of how do you score? How do you move through your pipeline? There’s a lot of insights that aren’t being captured.
So, that’s a great use case that you’re seeing some value added benefit with CLIPr that AI, and additionally with the hospitals and the medical universities and researchers alike, the notion is that a lot of teams are still using antiquated technology and they’re not extracting insights from these video moments.
I can hear as you’re sharing the excitement about your product roadmap, that you’re also looking at the data, the data labeling and the insights, and what’s shared from these organizations. I’m sure some of these might be new product features in the roadmap, from where CLIPr.ai is today to where it’s going, whether you see the continued growth, both with your clients and with the product.
We’ve gotten so much feedback. New things that are kind of coming out soon, things like CLIPr reactions, CLIPr comments, and these are actually kind of adding to the collaborative features of our CLIPr notes, and basically the reactions and the comments are going to allow for people to interact at the frame level within the video itself.
Whereas like right now, if you look at YouTube or Vimeo and you make a comment, you’re making a comment on the whole video, you’re not making a comment inside the video. So people are going to be able to anonymously interact off of CLIPr, within those moments, and then when they create an account, then at that point they can share it. They can upload their own content. So for us, like we get so much feedback around how they want things to be. For us, we’re really obsessed with customers and like making them happier. For us, like initially we’ve been spending a lot of our time on powering the event platforms and the event platforms are servicing many event organizers, and it really leverages a one to many ratio, and so in that part of the business.
We’ve actually also come out with the CLIPr open API, which actually allows for developers to integrate CLIPr embedded into their line of business applications so that people don’t need to have a context shift. So if they want to upgrade their video from YouTube or Vimeo to CLIPr, then at that point, the CLIPr video is sitting there and the event organizer can actually go and make the video available, then the end-user clicks on it. The event organizer actually can see what that person benefits from the video, what they end up not benefiting from it, how they’re actually smart skipping, because smart skip is a feature that lets you intelligently go from one topic to another topic. When people skip things, that’s actually value added information for an event organizer or a content owner who ends up understanding implicitly, which part of the content was more relevant and which part was less relevant.
So that when the up occurs you can actually go, looks like you are really interested in blah, blah, blah. At that point it’s a much more precise and actionable lead as opposed to: Hey, what did you think? Because in a way, like you ended up effectively knowing what they actually appreciated and what they didn’t appreciate.
So a good chunk of a roadmap is actually helping a line of business applications to better integrate CLIPr, whether it be an event platform. Whether it be a CRM platform, even in recruiting. Another use case that people are using CLIPr for is a new procedure that has evolved over the past year is that people are now recording interviews.
So, one company that we’re working with is using CLIPr to actually identify a hire and no hire moments. So we’ve all had those discussions where you’re evaluating a candidate and you’re like: I really want to hire this person because of X, Y, Z. And now that recruiter or a hiring manager can actually play X, Y, and Z. So the other hiring managers or other recruiters can actually see a higher moment and no higher moment, or actually see how that person answered that question and why that ended up being much better than how someone else answered it.
So you could see how people’s skills can actually improve based on referring to these recordings, which up until now were just recordings that were in people’s memories or jotted down in Evernote or on pen and paper or in an email. But now you can actually refer to the actual video moment, and it becomes a whole new vocabulary to refer to the CLIPr real, and it just becomes a new way of work, a new way of hiring, a new way of improving, surgical skills and all hands meetings, people are always announcing things, new hires, new products.
When a new company, new employees come on board, they can actually refer to previous all hands meetings, because interesting things were discussed. So this whole idea of like referential utility can, actually, be finally realized with CLIPr. Whereas like right now, it’s just been a concept. Part of the excitement for CLIPr, and everyone on my team, is that most people don’t know that this is possible and once they realize that it’s possible, it’s like: Oh my God, I can’t imagine life without CLIPr.
That’s kind of like, what we want to do is create these AJA! moments where they realize the power of the technology and they start using it. But we also realize that we’re instilling new behaviors and that also takes time. So with that we’re actually scaling the business through partners who already have relationships with enterprises, with the universities, with the event engines, because we can’t do it by yourself. We’re still a small startup, but growing rapidly.
Some of these events that you’ve mentioned, like all hands, when you’re a company, suddenly, you can have dozens of these all hands in a calendar year, and when a new hire comes in should they be watching many hours of these or go to those referential moments or those CLIPr reals, which can be so powerful there. I know, also, firsthand as having been a hiring manager at multiple companies and interviewing candidates is a very time consuming, but necessary process to make sure you’re effectively sourcing the right candidates though.
Of course not everyone can attend every interview and telling all of your team members: Hey, can you watch the 60 minutes replay of an interview, may not always be very efficient. So I do find that use case of here are these three stand out higher moments. When this candidate crushed this coding question or where they aced this technical portion or were spot on with this behavioral use case can be so value add.
So, I’m really looking forward to seeing where you take everything with the CLIPr.ai, the open API, with your comments, with your reactions, with all these expanding features, as you continue to navigate growth in 2021; and with your previous work with scaling up different moments at AWS computer vision and other video startups that you’ve seen this whole industry continue to grow and evolve. I want to hear some of your thoughts on trends, highlights and where you see just the industry as a whole going over the next few years.
With machine learning, in general, it’s all about the data and about engagement and interaction and training new models around the data. So, for me, it’s really exciting because we all know that with machine learning a lot of times people create things and they’re looking for problems to solve and they’ve got a solution. In our particular case, we’ve latched onto a very real problem that everyone has, so all the tools that we have ended up being applied to it, and they’re going to keep getting better and better and more and more efficient. So for us, the labor intensity that we have upfront is going to actually consistently go down because as we get more data to train against cross-platform, across all the different video types it’s going to get more and more efficient.
That dynamic that we’re experiencing, others are also experiencing as well within their own very specific use cases. So, in a way, I kind of feel like the overall trend is that technology is actually going to find more and more meaningful problems to solve, and, actually, we’ll be doing them even more effectively and more efficiently.
So, that’s like a top level trend, which is intuitive, because we all want things to be better, faster and cheaper, and what essentially is happening is that technology is actually allowing for that tri-factor to consistently occur across like all the areas where there actually are real problems.
If there isn’t a real problem, then time is being wasted. So, for me, tapping into the power of, in this case, AWS, because they are our backend, It allows us to do so much more and to do it so much more quickly. And it’s not just us. This ends up also being applicable with the Surer, with GCP and each of those different cloud providers. The fact that they are investing so much money in the technology is actually allowing us to actually do so much more than was ever possible.
When our investors take a look at what CLIPr has done in the past few months, They’re just shocked at how much has been done. Normally would have been like a multi-year thing and we did it in months, but we built it on top of years of work. So the fact that these things can go from concept to reality so quickly, It’s so awesome. The fact that we’re even focused on trying to save a bleeding hours, people would laugh at that ambition, but that’s just five saving 5% of information workers, 30 minutes a day, that’s not that hard. So we can actually think really big and actually make it happen. And so in general, you’re finding that people can be way more ambitious and actually can actually execute on it. At this point, you’re only kind of limited by imagination, cause the tools and the technology exist and the money is actually available too.
All of these things are really exciting for technologists and entrepreneurs to make meaningful stuff happen. That’s really kind of like what we’ve been focused on here at CLIPr.
That is so exciting to see with the hyperscalers, like AWS, that today there are so many services out there and the purpose of services are to build businesses and to, of course, augment either with the stable technology or the experimental technology for what will be the future of AI of natural language processing of emotion, detection of different technologies. I know that your team is working on firstname.lastname@example.org. Where do you see some of the additional progress that still needs to happen beyond the data?
This is kind of like where we’re actually looking for inspiration around like other ways for CLIPr to come into play and we’ve heard of things like, in telemedicine, people have EMR with electronic medical records, but now doctors are also recording these patient visits. So if they aren’t able to refer to it easily, it’s the same as not existing, like courtroom proceedings, like we all know of you’ll the person typing as a teller, I forgot the name of that person’s role, but like CLIPr is recording and can actually be used in courtrooms.
So for us, we’re actually seeking out really interesting use cases where CLIPr can have high value and high impact, and that’s areas where we could use help and inspiration for like, and the other thing too, is that we actually support 31 languages. So we’re, actually, looking to internationalize ourselves. So there’s parts of CLIPr that are fully automated, but there’s parts of it that actually require a human in a loop and some of the topic labeling aspects does require a human a loop, but we haven’t prioritized clipper topics outside of English.
We have prioritized German and French, but we need to know clients that actually need this and want this in order to kind of help grow that out, including, Korean, Japanese, and Chinese. These are things that we can already do, but then we also need to focus. Cause we don’t have unlimited resources. So,for us right now, it’s really kind of identifying these design wins, cause we all know that, in crossing the chasm right now our sweet spot is the early adopters and those early adopters are going to have a problem, and if we can help solve it, then they’ll use us, and I would like to find those early adopters that will, actually, work with us to identify these use cases, make it very successful for them.
So that when we kind of do our eight plus round series we can actually highlight to the world all of the progress and the impact that we’ve had across all of these different verticals across all these different segments, so that the followers and the masses can actually follow the early adopter lead. So for us, it’s really, right now, it’s really evangelism around CLIPr understanding what it can do and then helping us to figure out how we need to make it better, and as new features get uncovered with specific use cases that’s actually going to flow back into our video analytics and management platform so that everybody can benefit from it.
So for us, it’s like helping us find new problems to solve and leveraging the tools that we have for all sizes and in this case and users can come into CLIPr. But right now it’s primarily the event platforms and enterprises, and with CLIPr getting embedded into more and more things, we want our technology to be a new ingredient for as many videos as possible. We also want to do it in a very scalable way.
With your continued acceleration for your product led growth, you’ve recently also announced a new round of funding to continue accelerating the product and your scale-out. Can you share with us more about that today as well?
We’re really excited about this. Earlier in the year when we did our initial private beta and public beta in December. In January we were going to raise a seed round and ,basically, market fit takes a while, and in this case we, actually, have found a strategic partner that really resonates with CLIPr’s value proposition across all the segments that we are working in.
So with that, we’re really excited to announce that we’re closing, basically, a 5 million seed round with a strategic partner by the name of Poly (P- O- L- Y) many of you may be familiar with Poly in the context of Polycom speaker phones, which are pretty pervasive in the artifact once known as a conference room around the world.
But the other part of Poly includes Plantronics,which is their ticker tape on NASDAQ. So people also know of Plantronics through enterprise Bluetooth headsets, as well as call-center headsets, but remote working solutions in general. So, when we think about ideal strategic partners, Poly has been focused in the unified communications world as well, but very much from an endpoint standpoint.
So, whether it’s a speaker phone or whether it’s a headset they’ve been working in this unified communications way, cross-platform, they’re also one of the largest resellers for Zoom, Teams and Meet. So, they saw a natural fit with CLIPr because we also are doing next generation unified communications. But in this case, we’re focused on machine learning and we’re also cloud-based, and rather than just being an endpoint, we actually analyze the data, allow for referential utility, allow for collaboration and end up also allowing for monthly recurring revenue, because It’s just more strategic.
So I’m going to go from having one and a half salespeople to like 300 plus around the world, and instead of initially selling to enterprise and events friendlies, I’m going to be able to leverage a salesforce that’s already in 90% of the Fortune 500 and Global 500 around the world. So for us, it’s a huge scale multiplier, because it’s one thing to have a strategy of landing and expanding.
It’s another thing to have CLIPr bundled into offerings that are already being sold, bought and used with a really meaningful and credible enterprise channel partner. They also have channel partners. They also have their own partner ecosystem that ends up becoming a direct extension for CLIPr as well. So yeah, this is a really exciting moment for us to be able to like work with the strategic partner that’s going to help us grow, help us make our product better and help us get to market sooner.
Well, it sounds like the next year is going to be an exciting time for both product led growth and business led growth. For any listeners, checking out the show today, you can find more information about CLIPr. AI, video analysis and management platform and what it can do for your enterprise or virtual event by visiting www.clipr.ai. Humphrey Chen, the CEO and co-founder of CLIPr.ai. Thanks so much for joining us on the show.
Thank you so much.
Thank you for listening to this episode of the HumAin podcast. Did the episode measure up to your thoughts and ML and AI data science, developer tools and technical education? Share your thoughts with me at humainpodcast.com/contact. Remember to share this episode with a friend, subscribe and leave a review, and listen for more episodes of HumAin.