Solving New Business Challenges with Generative AI
In this episode, Michael Daehne, Founder and CEO of FlexPoint Consulting, talks with Shikhar Singh, Enterprise Solutions Lead at Copy.ai.
Shikhar describes how the team at Copy.ai has built and personally used generative AI tools not just to boost productivity but to solve new business challenges.
He invites us all to ask: if capacity weren't a constraint, what would your 2024 business plan look like? What initiatives would you put on the list? And how would you go about tackling these initiatives with the benefit of generative AI?
Learn more about Copy.ai at https://www.copy.ai/.
Teaser
Michael Daehne: If we had video on, we could call this like two bald guys in tech, but it’s not that, so we won’t call it that.
Opening
Producer: Welcome to another episode of Inflect.
In this episode, Michael Daehne, Founder and CEO of FlexPoint Consulting, talks with Shikhar Singh, Enterprise Solutions Lead at Copy.ai.
Shikhar describes how the team at Copy.ai has built and personally used generative AI [artificial intelligence] tools not just to boost productivity but to solve new business challenges.
He invites us all to ask: if capacity weren't a constraint, what would your 2024 business plan look like? What initiatives would you put on the list? And how would you go about tackling these initiatives with the benefit of generative AI?
We hope this episode provides some helpful examples in thinking through these questions.
Introduction
Michael Daehne: Hey Shikhar, welcome to Inflect. How are you?
Shikhar Singh: I’m doing awesome, Michael. How’s your day going?
MD: It’s going great. It’s been a great week, and this is a perfect way to end the week, talking about some cool, exciting things, specifically around AI. I’m really grateful that you took the time to do this today. You and I go back a long ways and worked together back in the day, and I think so highly of you and value your opinion and insights and forward-looking vision. So I’m really excited to have you on and talk about some of the stuff you’re doing and how that may be relevant for our clients.
SS: I’m stoked to share more of that but really likewise on all of those points. It was awesome to find an excuse to reconnect with you, Michael.
MD: Yeah, for sure. Hey, I know a lot about you, but for our listeners who may not, can you tell us a little bit more about yourself and your career background and certainly about Copy.ai.
SS: Yeah, I’d love to. I’ll start most recent to first, or more recent first here. I’m currently the Enterprise Solutions Lead at Copy.ai. What that means is that a big aspect of my role today is actually helping customers understand the capabilities of generative AI, how they can actually leverage it in order to hopefully do something similar to what we’re doing, but actually really impact the way that they’re really driving outcomes to the business.
Prior to this, I was actually head of solutions architecture at Moveworks, which was another AI company that was actually started in 2016. And prior to that, I spent eight years in consulting. So pretty much my entire career can be summed up as someone that has spent a lot of time working with enterprise customers, advising them on everything from strategy to what app they should build next. But most recently, I’ve spent a lot of time really helping customers understand the capabilities of AI and using those capabilities in a way that actually are helping to drive the outcomes they care about most.
MD: Yeah and, unlike me, you were wise enough and smart enough to save yourself and get out of consulting. So kudos to you on that front. I’m still on the dark side, I guess, as they say.
But yeah, really good background for our listeners about your journey and experiences.
Copy.ai Background
Michael Daehne: Leaning more into generative AI specifically, tell us a little bit about what Copy.ai is doing for your clients and then maybe we can build on that to broader enterprise use cases.
Shikhar Singh: Yeah, absolutely. So if you haven’t heard the story of Copy.ai yet, let me actually summarize why I decided to join this company, and honestly, one of the coolest things that I’ve gotten to experience in the entirety of my career.
We started building on the generative AI stack in 2020. We started off building actually on GPT-2. Since then, obviously, things have grown quite a bit, but I can really summarize our journey with three numbers. In 29 months, we’ve amassed 10 million users, and everything that we do is powered by 36 people. And that foundationally is not really a set of numbers that you hear together, especially when you think about really what it means to actually be able to deliver impacts for that many users. And so that was my first point of intrigue when considering what I wanted to do next.
But I think a key aspect of really what makes that possible is actually the core business that we’re in, which is leveraging the capabilities that we’ve learned to build the tools that we need that will allow us to scale our growth. And, in addition, really create a different work experience that can actually allow us to use AI in order to be able to focus on the things that matter most and automate a lot of the white-collar assembly line work, so to speak, that tends to get in the way of what we love.
MD: Yeah, what I love about that, Shikhar, and you mentioned this to me offline before today, but y’all are really kind of living and doing a lot of this internally to enable your own growth – I think that’s what you were alluding to – in the same way you’re trying to help your customers do it.
Examples of How Copy.ai Uses Generative AI
Michael Daehne: I know you mentioned some examples around webinars and sales prospecting, maybe give a couple of those examples of how you’re using AI internally.
Shikhar Singh: That’s a great call-out, Michael. There’s a lot of noise right now around the basic co-pilot application for AI. And sure, we’ve all heard great things that people were able to do on ChatGPT. What we’ve probably also learned is that people tend to be more successful when they’re able to break down a more complex task into a series of components. So, we’re basically doing that with our own platform that we’ve built primarily to automate a lot of things that I think really get in the way of focusing on the things that are most important.
To give you a few use cases, we run webinars weekly. It’s a key aspect of our growth strategy. Of course, we want to make sure that we’re spreading what people can do with our platform. At the same time, our entire performance marketing team is actually just two people. So, as you can imagine, there’s a lot of pain that goes into even running a webinar. First, you have to know what you’re going to talk about, then you have to market the webinar, you have to create a description, you have to create all of the social content and all the other promotional content. Then after the webinar is over, you actually have to do all of the content repurposing, create YouTube-ready descriptions, create snippet videos that can be shared on other channels. That entire process used to be basically an entire week.
With our workflows platform, what we’ve been able to do is actually create a series of AI steps that took that business process – all the things that we do pre-webinar, all the things we do post-webinar – and simply use our engine in order to be able to take a few bullet points and create every asset that we would need to market a webinar. Then, after the webinar, we just take a transcript or an audio file and actually produce all of the outputs that we would otherwise want to create: YouTube-ready descriptions and all of the content and repurposing.
The impact of that is, frankly, it just takes a couple of minutes. The folks on our performance marketing team are, of course, probably the happiest about that.
From a sales perspective, think about all of the work that goes into prospecting, figuring out who someone is, what they care about, why they might be interested in talking to you. This is, you know, honestly a fairly challenging job. Well, it turns out that if you can take the best practices that a typical sales development representative would leverage, we can actually use that same capability to, say, get content from a lead enrichment network or LinkedIn and write personalized sequences, so we’re not sending those generic emails that you’re probably deleting. And the list goes on and on.
Our approach to building on generative AI is actually focusing on the things that would otherwise require us to scale in a natural way and find a path to be able to automate that using our platform.
MD: I love those examples and certainly they resonate with me in terms of some of the outbound sales and marketing activities.
I think an interesting aspect of this, that I’m sure are you are feeling given your line of work, is AI and generative AI are not new. They’ve been around for a while, but they have certainly captured the hearts and minds or the attention in the last, call it six months of the world, more so than they had before. And so, there’s a lot of talk about it.
Transformational Opportunities with AI
Michael Daehne: I think what I’m finding with some of our clients and partners is folks understand it, they understand how AI and how generative AI can accelerate or automate specific tasks.
I think what’s interesting about what you’re describing and what y’all are doing more broadly is thinking about not just how do we improve each widget in the process, but how do we think about a broader process through a new lens? Can you talk to that a little bit more, about this idea of taking a step back and thinking about truly transforming an organization through these tools and tactics versus just making each little piece a little bit better?
Shikhar Singh: So when you zoom out and really think about what the opportunity is for AI for a business, I think there’s an obvious desire to just think about getting everything that you’re needing to get done today faster. And let’s be honest, we’ve probably all heard stories of the people that have seventeen jobs and are just writing code with ChatGPT until they get found out and fired. That’s probably not the best use case, though, albeit it’s probably a challenge for some organizations today.
It's not just about getting work done faster, right? It’s about actually thinking about the zero-to-one opportunities that present themselves. And I think marketing is actually a great example. For the last decade plus, every company that has been interested in solving problems that are near and dear to any marketer’s heart has been amassing mountains of data to better understand their users, better segment them, better understand how they can message to them, better understand what their behavior is, and ultimately tailoring an experience that is more likely for them to convert. Now, ultimately, after doing all that work, what do we do? We send them a generic email. Say we’re an e-commerce provider, we’re going to send them the same abandoned cart email that everyone else gets. We show them an ad, it’s the same ad that we show to an engineer… well, that’s maybe a little bit more hyper-specific, but you get the idea.
Our content is only as good as our segments. And our data, despite being at the user grain, doesn’t actually influence our outcomes.
And so, when you think about what the transformational opportunity with AI is, say, for a vertical like marketing, it’s actually just to have a one-to-one conversation with all of the people you’ve been aiming to have a one-to-one conversation with. What generative AI presents is really a solution to the last mile problem. That means that rather than just trying to get the same work done faster, if we can actually use the data siloes that we’ve been creating and pair them with the processes that are emblematic of our best creators, what we can actually do is create outcomes that are in line with everyone that’s involved, right, so leverage the data that we’ve collected to speak to an individual customer and actually influence new outcomes, better understand their feedback, aggregate that feedback at scale, and feed it back into the flywheel that we’ve created across our product or engineering teams.
There’s obviously a ton of other zero-to-one use cases that you could think of. For example, movie creation or audio creation or really a lot of the other sides or other domains that we might think of. But that foundationally gives us a mechanism to create transformative impact for an organization that goes well beyond just doing something a little bit faster.
MD: Yeah, I totally agree. And I think another way of putting it or thinking about it is not how do I use AI and automation only to cut costs and only to solve the productivity problem. Of course that’s important, like you said. But I think the more interesting, exciting question is how do I use these tools and capabilities to grow my top line, to reach a new market or new customers, or do things that, to your point, I haven’t been able to do for 10 or 20 or 50 years because of some capacity constraint, human or otherwise. How does this remove some of those barriers?
I think the leaders and organizations that are asking the question in that way, not just how do I use this to cut a little cost, those are the ones that are really going to thrive in this new era.
SS: 1000%. And that’s the conversation that we have with, or that I have with, prospects every single day. You know, really opening their eyes to think a little bit differently about what generative AI could be for their business. Because I think that there is, you know, obviously a natural desire to see it as, wow, this thing could honestly represent a huge cost-cutting potential for my organization and help me get a lot of the work that takes a lot of time done faster. There’s value in that velocity. But I think we’re missing the forest for the trees when we start to do that as a starting point.
MD: Yeah, for sure.
Reframing Ways of Thinking about AI
Michael Daehne: You mentioned your conversations with clients and prospects, and I know you are living this every day. What else do you think might be broken or suboptimal or – I don’t know the right way to put it – a little off in the way organizations might be thinking about AI? Are there other things you go, man, I wish more organizations would think about it in this way or through this lens?
Shifting from Cost-Cutting to Using Productive Capacity Better
Shikhar Singh: Yeah. So, let’s start with just one of the common tendencies that people have when they’re thinking about what AI can do for a company, is actually to frame it within this lens. Say, for example, I’m talking to a CIO. The CIO’s desire will probably be to think about it from a cost-cutting perspective, or that might be the natural starting point of the conversation.
Ultimately, the CIO is obviously important in making the decision, but there’s a wide variety of other people that are involved, right? Typically, a VP or some trusted advisors are going to be involved in actually helping to scope out use cases, be a part of the evaluation, assemble kind of a tiger team that can help them figure out how to make that investment best.
Now, when you frame it as a cost-cutting measure and you know that you’ve hired great leaders, great leaders want to protect their people. And that can sometimes make it really, really challenging to actually ensure that these evaluations are conducted in a way that is effectively going to produce the right outcomes for the business. And that’s why it’s, I think, so important for companies to think about the zero-to-one opportunity, as opposed to thinking about it as simply a cost-cutting measure. Because the way that we frame the value of AI and the potential of AI within an organization will drastically influence (1) whether or not these projects are successful, (2) what you even go about testing, and (3) the desire for the folks that are actually boots on the ground, that have the business processes, that you trust to actually conduct a lot of the work that you need to get done, from effectively being as transparent as they could be and as invested as it could be in those outcomes.
And rather, I think, a better framework to use is to think about all the things that you actually signed up to do when you decided to join a particular company. The automation potential and the obvious value of that automation isn’t that those things just are the only point to your life, right? The only point of your day-to-day work. But actually to ask the question, what would your day look like if those things didn’t take time anymore? What could you do with the excess capacity that you’ve created?
And this goes directly back to your point about making top-line revenue a bigger priority. Effectively, you could actually leverage the same productive capacity that you have today in order to be able to drive significantly more impact, from a business analyst all the way up to a VP, all the way up to an entire new product line or other key strategic projects. And that’s a way to reframe the way that, even just doing the same work faster could be, it could be as a mandate, a top-level mandate, for generative AI evaluations.
MD: Yeah, I love that. I love that you use the word impact or purpose, and I think that’s a really helpful lens for organizations and leaders to look at all this through, as well as yes, of course, we’re trying to drive shareholder value and drive profits and all that. But every organization, at least every good organization, has a purpose, a purpose in addition to or tied in to its core being. And so I think it’s really interesting to ask the question of, how could we use generative AI to better meet our purpose or to better scale our impact? And, of course, money will follow, right, for for-profit organizations, if you’re doing it well. I think that can help, to your point, help broaden the conversation so it’s not just a cost-cutting discussion.
SS: 100%.
Rethinking Tool Selection and Procurement
SS: And, you know, I think that one of the other things that I think is a bit broken, or at least probably will be very broken in the next, I don’t know, year or so, is: we’ve already seen more model releases and more startups born in 2023 alone than have been in the last five years. And what that means is that there’s going to be a massive selection of potential tools that teams have the option of adopting. And if you think about a typical procurement process or think about a typical evaluation cycle, there’s simply not enough time and there’s too much at stake to evaluate one tool for one team for one purpose. And that actually means that, in order to reap the most out of our investments, we have to get really good at either evaluating tools very, very quickly or actually looking for tools that effectively allow us to solve more than just one problem for one stakeholder.
Because I mean, I’ll be the first to say, that the genie is out of the bottle, right? All of your employees have likely used ChatGPT and may or may not have had successful experiences with it. But, at the end of the day, I think every employee in the world realizes that, you know, the notion that they’re going to be obsolete if they don’t get an understanding of how to use these tools is a critical aspect of really what’s motivating them to want to stay current, to want to keep up with these tools, to try what’s available And that’s probably something that’s evident in the demand for tools that, you know, a typical C-level executive is seeing across the business today.
So I think that the procurement process is just not really designed to be able to do that, and we really need to spend a lot of time thinking about what it means and based on how valuable it is to the business, how we want to actually conduct those evaluations in order to drive not just impact for the product team or a project that stays secret in the lab, but actually true operational value that puts these tools in the hands of creators, the hands of the people that need it most, and actually then evolves to be a fundamental fabric of the DNA of the business.
MD: Yeah, I think it’s such an astute observation you make about tools, the number of tools and tool selections and procurement processes and all that because, as you know, so many organizations spent decades building homegrown technologies. And then we were in this big age of enterprise software and SaaS applications where it was, hey, you’re going to get a subscription to this tool.
Fit-for-Use Tools
MD: Now we’re into and entering this world where you can’t just run your business with three big SaaS applications. You can’t just have one big CRM tool and a big ERP and call it a day. Organizations are going to have dozens or hundreds of tools, some smaller than others. And so figuring out a way – this is what we’re helping a lot of our clients do – figuring out a way to build a strategy and a culture that embraces using fit-for-use tools, where they make sense in the business, not being married to one or two big things. I think that’s going to be a big mindset shift. We’re already seeing it as a big mindset shift for a lot of organizations.
SS: Totally. And, Michael, we haven’t even entered the age of just-in-time applications. And I think you know a thing or two about how passionate I am about application development. But what I’m seeing in this space, due to the hypergrowth and hypercompetition of the model layer, as well as just the amount of innovation that’s happening in the AI space, is that there’s a world in the near future where the application that you need to solve your problem is something that doesn’t even exist yet, but will exist after you can describe it.
That’s kind of surreal, right? It kind of means that the value of software is in question. The value of VC is in question, but it also means that there is an immense value for being able to allow me to effectively have a platform that can create the tools that I need to solve my job, to solve my own problems on a day-to-day basis. And there doesn’t have to be a massive line of users around the corner that are waiting to adopt it.
That’s really what’s driven the growth of SaaS [software-as-a-service]. If you think about the use cases that have thrived there, everyone has a sales team, so they probably need Salesforce. Everyone has a website, so they probably need a CRM [rather, a CMS, content management system], whether it’s WordPress or AEM [Adobe Experience Manager], depending on the scale of your investments. But relying on these SaaS platforms, the Goliaths, to be able to solve and address each individual need, even if they are incorporating AI into the business, just means that you’re effectively ceding the AI strategy that you want for your business to whatever roadmaps they have. And ultimately when you get it, all of your competitors will get it.
So that’s why it’s important to think a little bit differently about really how AI fits into the business and: what is your generative AI strategy? What investments are you making and why are you making them?
MD: Yeah. I have loved this conversation. I think, as I mentioned earlier, I think most folks get it. Like, okay, AI is here. I need to figure out a way to leverage it in the enterprise. I don’t think there’s pushback on that idea. I think the challenge is how? Like, what’s the first step? How do I do this well? And I think you have provided a lot of great food for thought around how to maybe ask different questions and think about it a little bit differently in a way that will be very useful to our listeners and our clients.
SS: Thanks, Michael. And that’s just part of really my day-to-day. And honestly, I’d like to say that it’s just a part of my day job, but I can’t get this out of my head. This is the perfect match between all of the topics that I’ve kept up with or tried to keep up with over the years, whether it’s history, philosophy, tech, AI. These are all very important threads to me.
If Capacity Weren’t a Constraint, What Would You Do?
Shikhar Singh: But if I can summarize it as really one question that every leader should ask themselves. And this is really, you know, is not one that I think you can answer very quickly, but at least every time I’ve asked it, it’s tended to stump people, right? But it’s: if capacity weren't a constraint – if the people that you needed to get something done weren’t a constraint, whether you’re a leader, IC [individual contributor], or any other member within an organization – what would your 2024 business plan look like? What initiatives would you put on the list? How would you go about tackling these initiatives?
Because that’s the age that we’re in. We have infinite creativity at the power of our fingertips. And we don’t even have to pay the $20 a month for ChatGPT Plus to be able to leverage these technologies in order to be able to drive impact. It really is, in many ways, the great equalizer of our times. And it’s exciting to think about a world in which the to-do list that keeps us up at night and prevents us from getting started on anything actually becomes the greatest motivating factor for getting shit done.
Michael Daehne: I think that is a great note to end on. It’s an equalizer and a motivator for getting shit done. Couldn’t have said it any better myself.
Closing
Michael Daehne: So, Shikhar, thank you again so much for joining us. I really enjoyed the conversation.
Shikhar Singh: I enjoyed it as well, Michael. It’s been all my pleasure. The only thing I regret is that we don’t both have a giant plate of barbecue in front of us right now. So next time, maybe we’ll record this at Terry Black’s.
MD: Hey, I think that’s good.
On the last episode of the podcast, I gave a shout-out to the chocolate chip cookies at Tiny Boxwoods. And so I think it’s fitting that, on this episode, we give a shout-out to the brisket at Terry Black’s. And maybe one of these fine institutions will start sponsoring us, I don’t know.
SS: I would be back any day for really either of those two sponsorships. Those cookies are… I can’t get enough of them.
MD: Oh my gosh, you and be both. Well hey, thanks, Shikhar.
SS: Thanks so much, Michael.