How Jasper Grows: When AI-enabled Hammers Are Used To Hit The Right Nail
Digging into the generative AI platform race, and sharing insights and strategies on how AI-native startups can get big--stay big--and win amidst the AI hype cycle.
👋 Welcome to How They Grow, my newsletter’s main series. Bringing you in-depth analyses on the growth of popular companies, including their early strategies, current tactics, and actionable business-building lessons we can learn from them.
Hi, friends 👋
A few months ago I wrote what was probably my favorite—and most popular—deep dive. Perhaps that’s not too surprising, as it was on OpenAI, the company behind the fastest-growing product of all time: ChatGPT. Which, also, is probably the quickest product to reach the prestige of becoming a household name, as well as a verb (even my mom, who is certainly no prompt engineer, is ChatGPT’ing things.).
Back in Jan, when I published this first AI piece, getting through a conversation without the word AI (or, someone checking if you were using ChatGPT yet) was quite the feat. Everyone was scrambling to learn more, figuring out who and what to trust, and just trying to keep their head above the incredibly fast-rising tide of AI progress.
Social media, truly, was just like this. 👇
As you know all too well, nothing has changed.
And it won’t be changing anytime soon.
It’s taken nearly 70 years of research, experimentation, model training, progress, and failure in the AI field to bring us to this moment. Now, the rate of change is remarkable.
At first, things are slow. Then they happen quickly.
Like I called out in our OpenAI analysis, we’re probably still at the very bottom of a graph influenced by The Law of Accelerating Returns (AKA, compound interest for technological progress). 👇
And when thinking about this curve, one startup that’s certainly moving incredibly quickly along it is Jasper.
If you’re unfamiliar with Jasper, not to worry, we’ll be getting a lot more into it. But for context, they’re a B2B SaaS product that helps marketers create content (i.e. blogs, ebooks, ad scripts, website copy, etc) much faster using generative AI.
Now, the reason I picked Jasper for today’s analysis is because that dude telling you about all these new AI tools is right about one thing: there are indeed a shit ton of new apps coming out each week at a dizzying rate. 😵💫
Except, most of them will be gone soon. 🎻
Startups already face harrowing odds of success. Now throw some serious hype into the mix where people are hopping around like crazy sampling new tools, and there are (1) a lot of make-a-quick-buck apps coming out, and (2) people starting companies purely because of the interest in applying this new tech (AKA, solutions in search of problems).
The common thread between both of those cases is that many of these new products are missing fundamentals. And without fundamentals, building an enduringly valuable business becomes near impossible.
In other words, it might be easy to become big fairly quickly (in terms of revenue and user size), but staying big and relevant is the problem.
This surge in products is largely thanks to OpenAI’s accessible LLM (large language model), which has created a foundation that makes it trivially easy to launch an AI product.
To be sure, there’s nothing wrong with building on GPT, or whipping up something exciting, earning a shitload of money for it amidst the euphoria, and then fading away as people rush to the next thing. Lensa is a case-in-point. And even for the folks aiming to build long-term businesses, there will certainly be some consolidation around even the very best use cases. That's just how she goes. It’s the natural evolution of new markets.
But if you’re reading this newsletter, you’re probably more interested in building things that stick around and can survive consolidation.
Now, I’m not in the business of betting on which of these AI companies will succeed. But, I have been noodling on figuring out what will separate the winners and losers. So, today I’ll try to share with you what seemingly increases the odds of an AI-first product making it through.
And, let’s turn to one of the most successful startups in the space to help.
Jasper is an early adopter in using generative AI to solve a real user problem. And their approach brings us a fantastic example of a startup operating in the third layer of the AI stack (more on this shortly), and a product verticalizing—and commercializing—OpenAI’s GPT model for a niche.
They’re going hard after the $400B content marketing industry with the thesis that whoever understands the customer the most, has the most data, the best feedback loop, the broadest extensibility, and the most active community can build a moat that others won’t be able to beat. This is their bet in the generative AI platform race, which we’ll get into in a moment.
So far, it’s a seemingly solid strategy. Their growth trajectory in their two short years has been record-setting for early SaaS companies in both ARR ($75M+) and paying customer count (100K+).
Now, looking at a budding startup that was founded in 2021 and hit a $1.5B valuation within a year might seem like an overnight success story thanks to a lot of luck (and perhaps a premature growth analysis here). But in reality, it took Dave Rogenmoser and his two co-founders eight years and three pivots to hit their stride and find true product-market fit.
So, let’s see how they got there, and are continuing to drive Jasper’s impressive growth today. For folks thinking about (or busy) building an AI product, there’s no shortage of lessons here.
Here’s what we’ll be covering in our Jasper analysis:
The AI stack, and the generative platform race
The Road to Jasper: Lessons on pivoting, seeding a distribution advantage, and finding real market pull
From 0 to $75M ARR: Driving growth toward a $1.5B valuation and beyond
From MVP to a Platform that Crossed the Chasm: A lesson on community-led product development
From First-Mover to Defensible Generative AI Product: A lesson on AI moats at the application layer
Product-led growth, for AI-first startups: A lesson on how education is the best form of marketing
Small ask: If you enjoy today’s post and learn something new, I’d be incredibly grateful if you helped others discover my writing. Hitting ❤️ in the header, or sharing/restacking, are all a huge help. Thank you!
If you’re reading this in your email: click here to read the full post.
Before we get stuck into Jasper, here’s a quick primer on the AI stack, because…
For a fish to make sense, the ocean must be understood first.
— me 🙉
The AI stack, and the generative platform race
Generative AI is a branch of artificial intelligence focused on creating content (i.e. text, video, images, code, and audio) from a prompt. And it’s on track to become as fundamental of a technology in our lives as the internet, cell phones, email, and search.
Kai-Fu Lee said this: “I believe AI is going to change the world more than anything in the history of mankind. More than electricity.”
The gen AI space can be roughly broken down into three areas. Here’s a snippet from my OpenAI analysis describing them: 👇
The AI market structure
For the sake of this deep dive, and a usable-in-conversation understanding, I think the best way to explain how the AI industry is structured is this:
There are three layers.
The first layer is core platforms and infrastructure for AI.
The second (middle) layer is specialized AI models.
The third is the application layer — i.e. usable products/services built off L1 and L2.
Let’s go a bit deeper.
Layer 1 is these large language models (LLM), and the hosting services for them. It’s the foundational stuff that everybody else is going to get to enjoy. There will probably be just a few of these companies (i.e. an oligopoly), with companies like OpenAI. This is analogous to what cloud computing does (AWS, Microsoft Azure, Google Cloud). It’s super expensive to play that game at scale. So thankfully, they’ve taken care of “the server problem”, enabling builders like us to go and solve specific customer problems.
Layer 2 is the highly tuned, more compact, AI that will be built against the foundational models provided by layer 1 companies (via APIs). This is where there will be a ton of business opportunities on the table. It’s the specialization/verticalization of AI.
Layer 3 is all the commercial applications that will come from layer 1/2. These are the end-user-facing tools you and I will actually be using. Examples here are Midjourney, Jasper, and Github Copilot.
Here’s an excerpt from a fireside chat between Sam Altman and Reid Hoffman:
I think there will be a small handful of fundamental large models out there that other people build on. But right now what happens is a company makes a large language model (API enabled to build on top of it), and I think there will be a middle layer that becomes really important, where I’m skeptical of all of the startups that are trying to train their own models. I don’t think that’s going to keep going. But what I think will happen is there’ll be a whole new set of startups that take an existing very large model of the future and tune it.
I think there’ll be a lot of access provided to create the model for medicine, or using a computer, or a friend, or whatever. And those companies will create a lot of enduring value because they will have a special version of [the AI]. They won’t have to have created the base model, but they will have created something they can use just for themselves or share with others that has this unique data flywheel going that improves over time.
So I think there will be a lot of value created in that middle layer.
— Sam Altman, founder/CEO of OpenAI
And in a separate conversation between Reid Hoffman and Elad Gil, Reid said he was 100% certain that in the next 5 years, there will be “a co-pilot for every profession”, and that he thinks “there will be something [built on AI] for everything”. He added that he thinks that’s a generous estimate.
There are, in fact, two other massive layers in that stack though. This graphic below calls them out and also gives a neat visual summary of the above:
To the customer, Jasper is positioned at the very top: the end-user-facing, application layer.
However, while they started by plugging in and consuming OpenAI’s foundational base model (GTP-3) to get to market as quickly as possible, they’ve since been training and fine-tuning their own models on top of it, making Jasper both an L2 and L3 play.
This gives us an inkling that Jasper is not only becoming a co-pilot for content creation and a category-defining company in the marketing arena, but also one of the first startups outside of OpenAI to vertically integrate the value chain of AI.
That last piece is super important, here’s why:
In prior technology cycles, the conventional wisdom was that to build a large, independent company, you must own the end customer — whether that meant individual consumers or B2B buyers. It’s tempting to believe that the biggest companies in generative AI will also be end-user applications. So far, it’s not clear that’s the case.
To be sure, the growth of generative AI applications has been staggering, propelled by sheer novelty and a plethora of use cases. In fact, we’re aware of at least three product categories that have already exceeded $100 million of annualized revenue: image generation, copywriting [that’s Jasper], and code writing.
However, growth alone is not enough to build durable software companies. Critically, growth must be profitable — in the sense that users and customers, once they sign up, generate profits (high gross margins) and stick around for a long time (high retention). In the absence of strong technical differentiation, B2B and B2C apps drive long-term customer value through network effects, holding onto data, or building increasingly complex workflows.
In generative AI, those assumptions don’t necessarily hold true. Across app companies we’ve spoken with, there’s a wide range of gross margins — as high as 90% in a few cases but more often as low as 50-60%, driven largely by the cost of model inference. Top-of-funnel growth has been amazing, but it’s unclear if current customer acquisition strategies will be scalable — we’re already seeing paid acquisition efficacy and retention start to tail off. Many apps are also relatively undifferentiated since they rely on similar underlying AI models and haven’t discovered obvious network effects, or data/workflows, that are hard for competitors to duplicate.
So, it’s not yet obvious that selling end-user apps is the only, or even the best, path to building a sustainable generative AI business. Margins should improve as competition and efficiency in language models increases. Retention should increase as AI tourists leave the market. And there’s a strong argument to be made that vertically integrated apps have an advantage in driving differentiation.
— a16z
And as I wrote back in Jan: “To most people, for a long time, AI will likely be this magical thing that’s never fully understood. And it will be the companies that take that magic and make it easy and accessible for people to wield as part of their daily lives that will create a ton of enduring value. For example, a company like Jasper — who bottles the power of AI into a neat Web2 interface, making it approachable and useable for content marketers.”
This brings us to the end game these application-layer startups, like Jasper, are playing.
The generative AI platform race
The interface and workflow layers (L2 & L3) that are built on top of these LLMs are where there (1) will end up being a lot of consolidation, and (2) is the most potential to differentiate and use gen AI as a springboard (or trojan horse) to build platforms and network effects.
Similarly to when mobile started becoming mainstream, there wasn’t an influx of new phone brands coming out. Rather, it was the apps that expanded the hardware’s functions and use cases that came in droves. Including garbage fart apps, unnecessary alarm clocks, gimmicks, and, of course, real businesses that prevailed.
Now that the foundational layer is strong enough, and rapidly improving, we’ll keep seeing AI-first applications across all sorts of verticals flood the market. This space is where category winners emerge, more likely, as platforms. And remember, platforms win.
As an aside: Hype cycles like this—be it the internet, mobile, bitcoin, and now AI—often get bad reps. But mass interest is what ends up making technology flourish. You need excitement to get people to come and build products in nascent industries, for VCs to fuel them with capital, and in order to capture the attention of potential new users.
And those generative AI applications that make it through the fray and ultimately evolve into platforms will likely do so by pulling on a combination (if not all) of these 3 levers:
a) Controlling demand and deep customer focus within a vertical
The surest path to winning this game is by owning the customer.
As we’ll see with Jasper, their 6 years in the making bought them a huge distribution advantage: a deep understanding of—and founder access to—professional marketers.
At the end of the day, if your AI is inferior to a competitor but you solve the customer problem and JTDB in a better way, people don’t really give a shit.
Do you care that you’re getting a better/faster/more accurate solution because of what the latest technical white paper said, or do you just care that you’re getting a better product, period?
Exactly. Joe the copywriter couldn’t care less it took you a month to use the latest technical update in AI, he just wants to finish this blog post so he can get to his kid's soccer game.
Customer > technology, every time. And while Jasper is an AI-native company, they’re the perfect example of a startup that’s treating their AI as secondary — as a means to solving a problem in a better way.
Here’s an analogy: Don’t get too excited about the hammer and end up running around trying to find nails to smash in. Rather, focus on the right hole…and perhaps you’ll find a screwdriver is more apt. 🤔 Because right now, there’s a horde of generative AI–enabled hammers flooding Product Hunt, with founders caught in the hype cycle and forgetting the core lesson in company building: build something people desperately need.
Over time, the biggest share of revenue and profits will migrate to whoever owns the end customer and is using the right tool for the job, unless, there is some mix of proprietary and essential source material in the value chain. 👇
b) Having deep—and fine-tuned—AI by vertically integrating along the stack
Where there’s a stack, there’s an opportunity for savvy founders to vertically integrate along it.
As said, it’s easier than ever to get started by sourcing someone else’s model and finding a wedge at the tip of the stack. That’s where the customers are, and as we just noted, there is a huge advantage to getting a product into customers’ hands before anyone else.
But for the companies that are able to scale customers and revenue fast enough, it allows them to raise money and use it to build and fine-tune their own proprietary model based on an open-source code base. This comes with tradeoffs and benefits, but ultimately it’s a route to independence, more specialized AI for the needs of certain customers, and growing market share.
The likely rise of vertical end-to-end applications that include both a proprietary model and the application functionality is likely going to be the hallmark of winning platforms with staying power.
Looking at Jasper, we can see they’re already entering the middle layer by starting to build their own in-house AI. The main benefit of this—even beyond the de-risking of model reliance and avoiding an eventual model tax—is that they can be more customer-led since they have more control over how their specialized models work for this category’s needs. This is the main advantage Dave cites from vertically integrating down into L2. Of course, complexity comes with that, but it allows them to unbundle the massive marketing use case from ChatGPT.
Takeaway: Unless you’re an AI expert with an angle you know you can capitalize on, it’s probably best to start with an out-the-box solution, build some use cases around that, and then start to think about differentiation or diversifying as you grow.
c) Being widely accessible, deeply embedded, and feature-rich
Lastly, to win the platform race, AI-native startups need to be entrenched in customers’ lives.
They need to be easy to use and get value out of, embedded into important workflows, and complemented with tooling.
Because of the necessity to control demand, having a strong UX layer is table stakes. It’s also an opportunity, as it’s one of the few ways to differentiate from competitors and bake in some defensibility.
Just being a singular generative AI tool creates the risk of becoming a ‘feature’ of an incumbent platform, which has the advantage of an existing user base. Not a good fight to enter.
But, while singular generative AI tools may not be so defensible in the long term, they are still a great wedge when embedded into existing platforms or workflows and can build out their defensibility by (1) adding complementary features, and (2) ensuring they’re available wherever customers are working and need them.
As we’ll see with Jasper, they’re nailing all of these things. And on that note, let’s shift our focus to the team winning the race for the AI content marketing platform.
The Road to Jasper: Lessons on pivoting, seeding a distribution advantage, and finding real market pull
Jasper co-founders David Rogenmoser, Chris Hull, and JP Morgan started working on digital marketing startups together in 2014.
Although, their first venture together was far from an AI-native company driving content creation.
In fact, they didn’t even care what they did, as long as they made $6K a month to support their families. As Dave remembers it:
When we started in 2014, we just wanted to build a business together. We didn't really have any skills, and didn't really know anything. We were just like, “We don't want to go get a job, so let's go figure it out.” So, we started a marketing agency where we got clients and we would help them do Facebook ads, SEO, and build landing pages.
And this digital marketing agency didn’t do too badly, building it up to a decent ~$25K/month business within a year. Except, they were miserable and struggled to see how they’d scale it.
So, while sitting in a hot tub, they decided to pivot. Having built up a bit of a network and some expertise in the marketing world, they started selling marketing courses.
We basically productized everything that we had done successfully — and, the stuff we'd done unsuccessfully. One thing we were really great at was landing new clients and getting new campaigns out the door. And here, we were finally doing the marketing for ourselves. We were building great marketing funnels and were running our own ads at a much bigger scale than we had before.
That was an even deeper crash course. And those were skills that we now had in the tool belt to carry on to a lot of what makes Jasper successful today.
— Dave Rogenmoser
The trio did this for about 3 years, continuing to build their network and understanding of their customers. Then they pivoted for the second time, as they found that course businesses were great for learning but hard to scale. Wanting to go bigger, they turned to software.
If you’re a marketer, you may well have heard of the app that promises to drive website conversions by showing social proof. It’s called Proof.
And, you guessed it, these guys made it.
The idea was simple. On a landing page, or at checkout, drop in a little pop-up that shows real people who recently bought the product.
They whipped up a little prototype, threw it on their own course as an MVP, and instantly saw a lift of 48%. Knowing they were onto something, they quickly went to their marketing network to validate the idea. And as Dave says, it was far from a real product yet.
It was still this raw code. There was no app, no way to log in. We hosted a webinar and a couple hundred people show up to it. I'm pitching the whole thing, like “Here's what it's going to do, and here's all this magic”. It was a really great pitch. And then I said, "Hey, but it's not ready yet. But if you pay annually right now, if you pay $1,000 for the year, we'll get you first access whenever we do release it."
Everyone got pretty mad in the chat and was like, "I thought this was ready now, we want it." But then we had 80 people pay $1K on that webinar and that was the most money we'd ever made. We now had people that said, "Hey, not only am I telling you I want this, here's $1000 of mine to hold onto."
It was really validating before we had to go build this thing, knowing people really wanted it. And so we ended up building that over the next month and getting that out to people's hands.
And they certainly were the right people to take it to market. Y Combinator thought so too, funding them after they hit $175K MRR in 10 months.
Except, they found that Proof struggled to evolve beyond a feature into a platform. Facing high churn, a flood of competitors, and pricing pressure, after 18 months they decided it was time for another change.
Enter another hot tub and the fall of 2020.
We pulled back. I think this is a mistake founders make — when it's not working, they iterate too incrementally and they think, “Oh, maybe it's a little feature here. Maybe we need a different position to be better here.”
These weren't small changes [we needed to make]. These were bigger fundamental things that we needed to pull out of the forest.
Above the forest though, we looked around and were like, “All right, if we could do anything, what would we do?” And in that same month, we started digging more into AI and started looking more at GPT-3 and just realized how good these models were.
— Dave Rogenmoser
This text message from Dave to JP is where the idea for Jasper was seeded, marking their third pivot.
On January 15th, 2021, they launched Conversion.ai with this very customer-centric demo video.
Wanting to sound more AI-ish, they soon renamed to Jarvis. An ode to Iron Man. But, after getting a cease and desist from Marvel, they quickly changed it to Jasper. 💀
Jasper: Finding product-market fit by pivoting to AI-native SaaS
Very early on, Jasper had strong traction. And the way Dave describes their near-instant PMF is great:
I would read about product-market fit, and you’d hear people say, "You'll know when you have it." That was so frustrating to me because I was like, "That's not helpful. I don't know if we have it or not."
And then I realized I was missing the obvious. If you're frustrated by not knowing if you have product market fit, and you want the answer, you probably just don't have it. Because a month into Jasper, I was like, "Oh, that's what that meant. This is absolutely what it feels like to be just pulled by the market so much faster." For the first time, it felt like we were getting the benefit of something bigger than ourselves and we weren’t just grinding for every little dollar, or every little thing that we did.
Within a mere 9 months, they were doing $40M in ARR.
Probably the number one reason Jasper found such quick pull and validation from the market was that the founders spent 6 years leading up to it and figuring out the right thing to build for the people they knew best. 👇
Founder-market fit, and distribution advantages 📣
Across three pivots, each “failure” (as Dave calls them) gave the team the skills needed to build an AI product for their industry. As you probably noticed, each pivot parlayed on the previous idea, both (1) giving them compounding expertise in marketing and growth, and (2) growing the founders an audience of marketers to tap into.
They weren’t changing industries and hopping around willy-nilly. They remained focused on the same customer and applied their learnings to what they saw as the next level up each time.
This consistency—combined with a willingness to move on instead of banging their heads up against the wall—is what seeded them a few powerful distribution advantages.
And a distribution advantage is vital because getting your product in front of the right people efficiently—getting people to even know you exist—is increasingly what separates the haves from the have-nots. To break through the noise like Jasper did, you need to find a way to go directly to your early target audience cheaper and quicker than your competition.
As a startup, Lenny Rachitsky notes 7 distribution advantages.
Starting with a pre-existing audience
Developing a unique viral loop
Being first on an emerging platform
Having a remarkable story
Starting with pre-existing strategic relationships
Closing early strategic partnerships
Bringing extraordinary hustle
Dave, Chris, and JP bought a pre-existing audience, a remarkable story, and clearly hustle to the table. As Dave put it:
This was six years of working with marketing teams, helping marketers grow their business in a variety of ways, helping people write content better, helping people do copywriting better. I knew the customer deeply. I'm a marketer by trade. We had an email list of past customers built up and people that were following us.
Building a tool around AI was pretty straightforward for us. So I think we really had founder-market fit when we went into this.
If Dave and his team hadn’t grown up in the content marketing world, it’s unlikely they would’ve been the first to see the opportunity for how text-based gen AI could solve a huge problem for marketers: the need for near-infinite content.
This first-mover advantage got them one of the earliest partnerships with OpenAI, unlocking a 4th advantage: Being first on an emerging platform. Meaning, as soon as AI became the hot thing after the launch of ChatGPT, Jasper was ready for the spike in demand.
🛠️ What you can do with this: For starters—whether you’re a PM or a founder—choosing a market and customer set that you’re familiar with makes it a lot easier not just to understand who you’re building for, but also to really care about their problems. And for the founders, the value of building up an audience/community can’t be overstated. That could be as simple as starting a Substack, writing a newsletter, and building a network over time. You could start today.
Beyond founder-market fit and these distribution advantages, both the focus of their MVP and catching the right wave at the right time played important roles in their path to PMF.
Keeping their MVP focused on a narrow audience and use case 🔬
Jasper found rapid PMF by productizing GPT-3 for the serious marketers within Dave’s existing network.
By abstracting away the bare metal usage of OpenAI and reselling it with a pretty UX and user-friendly templates for copywriting, they were able to get something to market in less than 30 days.
This text-based tool—initially positioned to help with just writing Facebook/Google ads and landing-page copy— became their wedge.
Thus, a narrow focus on an acute problem that marketers face made Jasper super easy to explain to folks in demos, and as people flurried to try it out, they rolled out more templates and use cases.
🛠️ What you can do with this: Start very niche. And whatever you’re thinking, go even more narrow. The more specific the customer and your initial JTDB is, the better you can make your MVP appeal to them and the faster you can get them using it. This allows you to learn, iterate, and use this first target customer as a wedge into the larger market you’re after.
Catching the seismic wave of Generative AI 🏄♂️
Having a why now is super important.
How does the adage go? Timing is everything?
Some why now’s are better than others. Ideally, it’s a huge change in the world. A seismic wave.
If you remember much from school geography, seismic waves are those massive elastic movements in the earth that come from earthquakes, explosions, or other mega shifts.
In startup world, they’re long-term, transformative changes that shape markets. The big waves worth catching. The macro currents that create massive opportunities for folks like us.
On the technology front, AI is the clearest wave right now. And it creates a very strong why now for Jasper with two big advantages.
The ability to deliver a 10X better experience through (1) the ubiquity of a new technology, and (2) newly available data and APIs via OpenAI.
The moment ChatGPT proved that AI worked to the masses and ignited a tremendous conversation, it created a new untapped market need because people’s beliefs around AI changed. Therefore, so did their behavior and motivations. This supercharged demand.
Plus, the more recent a big change is (like AI), the stronger the current will be. There’s no denying that Jasper getting on early is bringing them a huge advantage in the (albeit still nascent) platform race.
🛠️ What you can do with this: Lift your head, and look out for big technological, behavioral, cultural, or regulatory changes. When you see them, ask yourself, “What hard problems does this change make easier to solve?”. Timing actually isn’t everything, but a solid reason why moving on your idea now matters sure helps.
From 0 to $75M ARR: Driving growth toward a $1.5B valuation and beyond
Okay, this is how we’ll break down the components of Jasper’s growth.
From MVP to a Platform that Crossed the Chasm: A lesson on community-led product development
From First-Mover to Defensible Generative AI Product: A lesson on AI moats at the application layer
Product-led Growth for AI-first Startups: A Lesson on How Education is The Best Form of Marketing
Loving the energy. 😎
From MVP to a Platform that Crossed the Chasm: A lesson on community-led product development
If you’ve read Geoffrey Moore’s iconic business book, “Crossing The Chasm: Marketing and Selling High-Tech Products to Mainstream Customers”, you’ll be familiar with this visual:
It’s all about what Geoffrey calls chasms: gaps between different types of customers within a market.
On the far left, there’s the smaller chasm. This first market hurdle is where a new product has to translate its idea into something beneficial and usable to people beyond a small group of die-hard tech enthusiasts — known as innovators. These are the people that will just try just about anything as long as it has something novel to it and they like what it stands for. They don’t care for bugs and will tolerate all sorts of product issues that the rest of the market won’t.
The key to getting beyond the enthusiasts and winning over a visionary [Early Adopters] is to show that the new technology enables some strategic leap forward, something never before possible, which has an intrinsic value and appeal to the non-technologist.
― Geoffrey A. Moore, Crossing the Chasm: Marketing and Selling Disruptive Products to Mainstream Customers
Then there is the big scary chasm. This is the main bridge the book focuses on, and it’s where startups often die in their journey going mainstream, as they fail to move between two groups:
Early market adopters and insiders → “Visionaries. The people who are quick to appreciate the benefits of a new product as long as it has utility and works.
Mainstream market → “Pragmatists”. The people who want to benefit from new tech but don’t want to “experience” the growing pains of it.
Trying to cross the chasm without taking a niche market approach is like trying to light a fire without kindling.
— Geoffrey A. Moore
In other words, the key to crossing the chasm is positioning and finding—and securing—a “beachhead” in a mainstream market. (Learn more about this)
As I described beachheads in How Canva Grows:
The strategy can be summarized like this:
Select your beachhead (Who is our target customer?): The idea is not to focus on a target market or target segment, but rather on a target customer you can access — AKA your High Expectation Customer (HXC). This is how you keep it super narrow and specific. A great example of a company we looked at that did this was Superhuman. You can learn more about how to tactically find your HXF here.
And an important note…bigger niches are almost never better than smaller ones. According to Geoffrey Moore, “The only time when a niche is too small is if it’s too small to generate half of next year’s sales”.
Get your product offer right for them (Why are they selecting our product over the competition?): The key difference between early and mainstream markets is that the former are willing to take responsibility for piecing together the whole product…the latter is not.
What you want to do is (1) get your value proposition right, and (2) develop a whole product, which solves all problems for a niche, then go from there. It’s about being laser-focused on the customer and the product they need.
Nail your distribution (What channels are we going to use to reach these customers?): Figure out the best way to reach this segment. Find out where they’re hanging out (think online and offline) and meet them where they are.
Push into your next wave of target customers (How does this initial customer group help us get to the rest of the market?): Once you’ve got your foot in the door (AKA your wedge), use that leverage and push onwards.
Applying this classic military theory from WW2 to this Texas startup, we can see Jasper is breaking into a more mainstream market by:
First, starting with a very niche audience: Dave’s immediate network of pro marketers
Then, launching a very scrappy MVP (conversion.ai) for these innovators and early adopters they had relationships with.
Next, formalizing a small and intimate community with them on FB. From here, leaning heavily into community-led product development, helping them figure out what Jasper needed to be. 🪄
With an improved product, starting to acquire the longer tail of hobbyists, small-mid size marketing agencies, and startups (i.e. going beyond Dave’s network).
Bringing these new customers into the community, and learning from them to figure out what the next level of the product should be for larger agencies and enterprise brands.
Breaking into the next tier of more mainstream customers with a whole product experience and momentum of social proof from 100K customers. Today, Jasper is reeling in big fish like Logitech, Volvo, Harper Collins, Zillow, Airbnb, HubSpot, and Amplitude.
After spending a lot of time looking at Jasper, I believe the most important aspect of this land-and-expand approach is the third point there: community. 🪄
So, let’s go deeper on that.
Leveraging the community to build a complete product offering
argues how community-led product development wins. He posits three reasons why:The customer is one of the most important people on the team.
You can’t build empathy for the customer by only relying on secondary sources, you have to immerse yourself in their problems by talking to them directly.
You should build a community to have daily interactions with customers about the product and their lives — it helps you build better products.
In other words, having a community is an excellent opportunity to interact with your customers, build relationships and brand advocacy, as well as build a very tight feedback loop with your end users. It plugs you into a constant source of innovation and ensures you’re aware of evolving behaviors and need.
From the get-go, Dave and the Jasper team almost solely relied on community-led product development to figure out what to build. Here’s how he describes it:
At first, we guessed it. We just said, "Hey, we don't have time to do a ton of user interviews. We just want to get this into the hands of the market ASAP and then let them pull that out of us."
There were a lot of things that were right, and a lot of things that were wrong about that. Our MVP didn't have almost any features or functionality. It was just the very core of the app. And we're like, “Let’s get it in people's hands right now”.
Then we heavily engaged. We had a community within our Facebook group from early on and I was just hanging out in there all day, every day, talking to people: "How are you using it? What do you guys need? What's bad about it? Where is it annoying?" And then we’d just try to ship something new every single day in a very lightweight way.
Speed was all we were trying to do. And just knowing that if we can just listen to them, build something that they say that day, this is going to work. That iteration cadence was so powerful. It also builds a lot of trust with your customers because they just think, “OK, they're listening and even if it's not what I want it to be right now, I've got confidence that it's going to get there.”
Because of this tight-knit communication with the community, they quickly realized in a matter of a few weeks that their first guess at a value proposition around conversions (AKA performance marketing copy) wasn’t what people were talking most about. Rather, people wanted longer-from stuff, like emails, blogs, social media posts, etc.
This led to a nimble repositioning and refinement of their value proposition. Now, thanks to the community, they have a product roadmap that addresses the pain points felt by their two major customer types – individual creators and larger enterprises. Namely:
For individual creators (like me), Jasper helps solve the problem of writer’s block.
And for larger enterprises, they help scale their content marketing efforts by empowering marketing teams to write copy much faster while maintaining consistent messaging and their brand’s unique voice. This allows marketers to focus more of their time on bigger-picture content strategy and initiatives.
And Jasper’s thriving community, and approach to building alongside them as an extension of their own team, is certainly a moat. As Kyle Poyar wrote:
Software companies used to differentiate on the basis of product features. With AI products, it’s safe to say that community is becoming the new moat (along with proprietary data). More robust communities draw in new users, ensure those users are equipped for success, and thereby fuel even more product adoption.
🛠️ What you can do with this: I wrote a bit on this a few weeks ago. If you want to go deeper into what building a community looks like, first check out “A founder’s step-by-step guide to getting your first 1K community members”, then look at this post on using the community for product feedback.
While on the note of moats…
From First-Mover to Defensible Generative AI Product: A lesson on AI moats at the application layer
While having the biggest and most engaged community—and subsequently a strong and authoritative brand— is a very strong defense here, there are two other important moats to expand on:
Data feedback loops around a proprietary middle layer, driving customer personalization
Extensibility, and breaking away beyond text generation into a SaaS platform
Let’s take a look, and then tie it together with Jasper’s defensibility flywheel.
Domain-Specific Model Advantage: Data, and the feedback loop moat
Where the last generation of creator tools like Figma ($20B) and Canva ($26B) built moats by converting low barriers to adoption into network and lock-in effects, gen AI apps will find moats that come from being deeply embedded into workflows and converting feedback data into finely-tuned custom AI models trained on customer data.
Jasper's existing lead as a content creator tool for marketers gives them access to more proprietary user input and engagement data, enabling Jasper to (1) store that knowledge in their own models, (2) produce the most customized AI that works super well for each of their customers’ use cases, and (3) improve Jasper’s domain-specific utility for their network of users.
In other words, each time professional marketers use Jasper—both adding different prompts and selecting outputs—they fuel the middle-layer feedback loop. Jasper learns more about each customer’s preferences, brand voice, and brand style, making Jasper a better and more reliable fit for each customer the more they use it. And because the AI has this valuable context and memory about each user, it makes on-brand writing exponentially easier — creating a neat lock-in effect.
Similarly to if you have a good therapist who knows all about you and can instantly give you tailored advice, there’s a switching cost. This is Jasper’s middle layer moat.
🛠️ What you can do with this: If you’re building an AI-first product and renting a foundational model, once you have real users and have validated that they actually want what you’re making, start considering how to weave yourself into the middle layer. Finding defensibility will be tough if you’re always reliant on a third party.
Note: You don’t have to do your own model training alone. Jasper partnered with Cerebras for help.
Extensibility, and breaking away beyond text generation into a SaaS platform
The other thing that we didn't see a lot of other companies doing was combining that on-brand AI assistance with the extendibility of being able to go wherever you wanted to create, and call up Jasper with just a keystroke. Our extensions and our forthcoming API are things that enable you to have your AI not tied so much to the tools that you use, but tied to you as a user and as a business, and available wherever you create.
You'll see more and more of that in our marketing and the way we talk about ourselves. But in another year there may be other companies that have caught up with that feature differentiation. And we'll need to look at that again. So you need to build on feature differentiation, marketing and brand differentiation, and community at the same time because it's the combination of the three that will help you go through those growth stages and not be overly reliant on any one area.
— Dave Rogenmoser
Think of extensibility like Grammarly ($13B). You don’t need to be on Grammarly to use it. Rather, it plugs into your browser and favorite writing apps, adding value wherever you need it.
For Jasper, meeting customers where they are is huge, because it captures more usage and engagement throughout a marketer’s day, allowing for a bigger and wider feedback loop.
Through Jasper Everywhere—their new browser extension—Jasper is plugging into different products and enterprise workflows (like Google Docs, Gmail, Salesforce Notion, HubSpot, Shopify, social platforms, and CMSs), making Jasper a deeply embedded part of their customers’ lives. This extension is their first step towards a fully unified AI experience widget, and Dave expects 95% of future usage to happen inside it, vs their standalone web app.
To that end, Jasper also recently acquired the Australian-based AI company, Outwrite (a browser and platform extension similar to Grammarly), with over a million global users. This acquisition accelerates Jasper's vision to bring generative AI assistance to creators wherever they work.
Becoming medium agnostic, and enterprise-friendly
Jasper’s go-to-market strategy was honing in on text generation for AI and doing that excellently. However, marketers don’t just operate in the world of writing. They need things like images, video, audio, etc.
In other words, Jasper’s customers are not constrained by a medium. And right now, every time they need, say, images for a blog Jasper drafted, they leave the platform.
Jasper wants to seal off this leakage.
So, towards becoming a whole product solution for their customers, Jasper is expanding beyond text. The most obvious, and easiest, next move in image generation given (1) the technology is working really well there, and (2) it’s the next biggest use case for marketers.
Recently launching Jasper Art, they’re bottling the magic of image gen AI which you could find on DALL-E or Midjourney into Jasper, within a much more user-friendly UX.
If you’ve ever used Midjourney, you’ll know you have to go to Discord, and kind of figure out how to even initiate a prompt for something. Also, you wouldn’t know to append things like mood, style, and medium, to your prompt unless you’ve seen examples. And then your creations are also just shared in a public chat room.
AKA, there’s just a lot to figure out and it’s not a very user-friendly product. Jasper solves for this by making it much more accessible. Plus, having this medium within Jasper adds another layer of feedback data. Each prompt generates 4 options, and as you make selections, their middle layer learns what you like and can more consistently create accurate stuff.
In the long term, Dave believes Jasper will become fully medium agnostic, adding audio and video generation to their offerings.
One on-brand content creation platform to rule them all.
This gives Jasper more coverage across the users’ full Job-To-Be-Done (JTBD), creating more stickiness and defensibility.
Aligned with this goal of being a full product, there’s also a scaffolding need: all the boring SaaS 1.0 stuff that is table stakes for a B2B customer. Think permissions, security, data privacy, and account settings. This is the less sexy stuff in the world of AI, but it’s also all the other things that a startup needs to actually complete the full JTBD for larger companies.
Plus, Jasper is laying in helpful SaaS features like team collaboration, tailored billing and SLAs for enterprise, a template library, plagiarism checkers, user management, SEO scoring, SOC2 Compliance, account managers, analytics, and more.
This makes them an obvious choice for more needy enterprises like Intel, the MAYO Clinic, and Experian.
Ultimately, both the productization and enterprisification (is that a word?) of AI will create massive businesses.
🛠️ What you can do with this: First, we know that solving for something that can primarily be done within generic LLMs (i.e. just using ChatGPT), is not durable. But neither is delivering a great standalone feature. If your AI-first JTDB stays too narrow, it will lack the basic scaffolding needed to build something that's ultimately defensible and has workflows that gather proprietary outcomes data.
Use first principles to build towards a whole product solution.
Tying it all together: Jasper’s defensibility flywheel
Here’s a rundown:
Value-driving product: The Jasper platform solves a range of use cases for content marketers, helping them get their job done quicker.
More users: As Jasper strengthens their brand and solves user problems better, more marketers come to Jasper and sign-up. Social proof and the community drive word-of-mouth growth, in turn, making the community even bigger.
More user interaction: More people use Jasper more frequently, and as Jasper integrates into new apps and workflows, their surface area for interaction capturing gets bigger.
More proprietary/active data: More usage means more prompts being added and more outputs being accepted, rejected, or tweaked. This helps Jasper learn.
Fine-tuned middle-layer model: Jasper uses the proprietary data from their customers as a feature extractor to train and fine-tune their own systems. This creates a lock-in effect (remember the therapist?).
More personalized (on-brand) AI experience: A better model makes Jasper’s content generator better, which makes outputs more consistent to meet and exceed users’ expectations. Specialized > generalized.
Superior user experience: Because Jasper creates more relevant content, users save time and meet their content goals quicker. So, they spend more time and engage further with Jasper, thus keeping the flywheel rolling. Combined with this is a roadmap powered by an active community, layering in more SaaS platform value.
Value-driving product: More engagement is a proxy for a customer-centric product creating a lot of value. The cycle starts over again, and the flywheel accelerates.
This is how Jasper keeps digging a deeper moat. ⛏️ 🏰
Product-led growth, for AI-first startups: A lesson on how education is the best form of marketing
Kyle Poyar raises a great question: “What does product-led growth mean in the age of generative AI? How do you convince folks to try a product that they fear could take their job someday? How should AI products be priced?”
Despite the buzz around AI, most people are still exploring it for the first time. Jasper’s challenge, as Meghan Keaney (Jasper’s VP of Marketing) put it, is:
…Educating the market that this kind of technology exists, making it incredibly simple for customers to find real value upon first use, and ensuring they understand why partnering with AI in the long term will make them more efficient.
That’s a challenge all application-layer startups will have to deal with. And the more enterprise-like the customer, the more concerns they’ll have about plopping their team and data into a startup that could be sending their data to OpenAI (thus, Microsoft).
Solving this education problem is not a straightforward feat, especially considering that for the majority of their customers, Jasper’s GTM motion is bottom-up, product-led (AKA self-serve sign-up). That means no sales or customer support folks talking you through the decision. Jasper’s product—the landing pages, sign-up flow, emails, the actual platform, etc—needs to convince people to sign up and whip out their cards for at least $40-$50 a month, per person.
Well, Jasper is clearly doing something right, given their impressive acquisition of over 100K paying customers in less than 2 years.
Jasper’s approach is a hybrid of PLG and sales-led. Remember, they break their customers into two segments: individual creators/SMEs and enterprises.
Enterprises get the sales folks to help them along. So, let’s focus on the first cohort.
On the PLG side, the top-of-funnel acquisition comes largely from paid social media, community-led growth, affiliate marketing, and free word of mouth from happy customers.
Generally, investing too heavily in paid acquisition early on isn’t the wisest move, however, Jasper found product-market fit very quickly. And I think ramping up user growth with paid spend is okay once you know you have PMF (before, you could be wasting money pouring water into a leaky bucket). Also, the paid growth engine fits into Jasper’s flywheel of attracting the most customers in order to build the strongest moat before the team over at Copy.ai does.
Then, once they’ve brought in the traffic, Jasper converts users with a free trial. This aims to convert free users to paid through a stick approach to the decision (i.e. don’t lose out on what you’re already doing). Once they’ve signed up and onboarded, users get a personalized template gallery based on some of the segmentation questions asked as they joined.
This flow from sign-up into templates is an excellent PLG strategy. It works great for Notion (read the Notion deep dive), and it’s definitely making the education and product discovery piece much easier for Jasper’s users. Instead of noodling over what your prompt should be (i.e., how to ask for what you’re asking for), they remove friction and, as Meghan says, “the magic moment is instantaneous”.
From there, the game is all about continued education and moving people along the product discovery, and value, curve. 👇
The more that people know how to use AI and integrate it as part of their writing process, the more likely they are to pay beyond the free trial, get the most mileage out of the product, and tell their friends. And beyond templates, Jasper uses their Bootcamp, Jumpstart (a collection of courses), community, live Q&A webinar-style calls, and a content-driven growth strategy.
Let’s go a tad deeper on that last piece, because what would a content marketing company be doing if they weren’t using content and SEO as a growth engine? AKA, what is it they say about skinny chefs? 🤔
While a mildly insulting cliche, there’s some truth to it in business. Luckily, Jasper eats their own dog food by using their own AI platform to create content. This (1) increases the efficiency of their own content marketers in building an SEO growth engine, and (2) helps them figure out which middle-layer tools and human + AI interactions work best for their users.
Jasper’s content strategy is editorially generated, SEO-optimized (learn more), and focused on product-led content based on their key personas and use cases. Seemingly, it’s broken up into four phases:
Phase 1: Drive relevant, top-of-funnel traffic.
Phase 2: Produce content that captures visitors further along in their buying journey using product-led content.
Phase 3: Capture searchers looking for specific AI tools and use cases that Jasper solves for.
Phase 4: Insert Jasper into the product evaluation process by capturing keywords related to competitors.
Each phase moves the focus closer and closer to product education.
At the end of the day, Jasper (like platforms often do) has a low floor and a high ceiling, all for the same price. And this education layer is how they inch users upwards to squeeze out the most value from it.
And folks, that concludes our Jasper analysis. 🫡
Personally, after studying everything they’re doing and learning more about the market, I feel I have a better sense of who will—and how to— win the generative AI platform race. Yes, there’s a lot of noise and eagerness in this hype cycle. And yes, it does make differentiation and defensibility tougher for newcomers who just need to get to market. But, I’ll leave you today with this recent thought by Anthony Pompliano:
The hype cycle is a necessary part of building out new technologies and industries. The capital shows up, but you need to be very cautious of how you choose to participate. Things will become overvalued incredibly quickly — don’t be the person buying at the top of a market.
I have no clue if the AI bubble is going to peak this week, next week, next month, or next year. Timing markets is a fools game. But I do think it is important to identify bubbles as they form. People will make a lot of money through the full hype cycle, and thankfully real products and services will be built, but you have to be careful that you don’t follow the herd into a losing proposition.
Contrary to popular belief, we need more hype cycles. That would be a sign that innovation and new technologies are coming to market. It also means that billions of dollars will trade hands, which will print massive winners and losers. Frankly, this is a story as old as time. Learn from history and try to avoid some of the mistakes that others already made.
That’s all, folks.
If you enjoyed this post, I’d be super grateful if you gave it a like, share—and of course—subscribe if this was your first time reading.
Until next time.
— Jaryd ✌️
This is a great post - I think my big question is whether they actually have strong retention or are growing a leaky bucket.
Conteúdo incrível e surpreendente!