The first batch of AIGC unicorns are already breaking up

Source: Silicon-based Institute, Author: Chen Bin, Editor: Boss Dai

In March this year, Zhu Xiaohu of GSR Ventures was invited to attend an industry conference. The main tone of the event was to "boost confidence", but Zhu Xiaohu poured cold water on the spot [1] ：

"I said before that the spring of enterprise services still needs to wait for 5-10 years; but after ChatGPT comes out, the cold winter may be endless. Please give up financing illusions in the next two to three years."

To support this point of view, Zhu Xiaohu mentioned an American start-up company called Jasper AI. Jasper AI is a SaaS company for advertising marketers, self-media bloggers and other groups, mainly providing copywriting generation services. Its core product uses the API interface of GPT-3, which can use AI to generate various marketing copywriting.

The company was established in 2021 and reached its peak when it debuted. In the first year, its revenue exceeded 40 million U.S. dollars, and it doubled to 80 million U.S. dollars in 2022 (ARR annual recurring income).

In the most glorious two years, VCs flocked to the company, the media rushed to report, and the company founder Dave Rogenmoser (Dave Rogenmoser) also became a new star in the AI industry.

But until November 30, 2022, when ChatGPT debuted, the situation suddenly took a turn for the worse.

**ChatGPT's powerful copywriting ability makes people suddenly realize that the "official" seems to be able to easily beat the "secondary creator" to death. **Jasper's products are similar to the "refurbished version of GPT-3". Facing the ChatGPT built by OpenAI, there is almost no moat.

Therefore, Zhu Xiaohu believes that [1] , the AIGC unicorn with a valuation of US$1.5 billion may "return to zero soon."

Unexpectedly, Zhu Xiaohu's words became a prophecy. Four months after his "bearish" remarks, Jasper founder Dave Loganmoser announced layoffs and reshaping the team on LinkedIn.

In fact, Jasper's "no moat" has been the consensus of Silicon Valley at the beginning of the year, but no one thought that the predicament would come so quickly. Of course, Jasper still has a lot of cash in its account (125 million US dollars was raised last year), and it is still a long way from "returning to zero", but there is no doubt that Jasper has broken a basin of cold water for AIGC's application layer entrepreneurship.

Regardless of the final outcome, as the first batch of AIGC unicorns to eat crabs, Jasper is destined to be a subject of repeated research and chewing.

The favored person

As we all know, AI is a field full of personal heroism: a certain bold move or whim may start an era.

For example, in 2012, Geoffrey Hinton, a professor at the University of Toronto, made a breakthrough in image recognition and almost single-handedly opened the golden age of AI; It has greatly promoted the progress of AI painting, and successfully made art students around the world start to worry about unemployment.

In 2021, Cade Metz, the editor-in-chief of The New York Times, published a book recording the history of AI development, titled "Genius Maker". However, Jasper, also founded in 2021, is an exception: the company was founded by a group of grassroots rather than technical geniuses. CEO Dave is a marketing background, and only one of the three founders knows a little bit of programming.

**The biggest contributor to Jasper's sudden success is actually Laman's "lucky talent". **

The three founders, including Dave, are serial entrepreneurs and started going to sea as early as 2015. Looking at their past projects, they basically have nothing to do with AI. And unlike OpenAI, which is "struggling to realize AGI", Dave's original intention of starting a business is actually quite "no dream"-just because he doesn't want to be a social animal from nine to five.

The trio's first company, called Payfunnels, essentially made money by "posing as marketing experts." The founder, Dave, hired people to pull orders, and picked out those small businesses that hadn’t seen much in the world, and made a lot of hype; after receiving the money, he took out half of it to buy the services of a genuine marketing company, and the other half went into his pocket .

In the days of being a "middleman", Dave stole a lot of marketing routines and turned himself into a "marketing mentor", relying on knowledge to pay for cash.

Until now, he is just an ordinary little boss who is unknown to the public, and investors don't even pay a second glance. However, fate seems to favor these "big fools" and insists on pushing them to the center of the stage of the times.

In order to assist the lecture, Dave's partner wrote a marketing plug-in, which can display in real time what products other people ordered. Unexpectedly, the conversion rate was as high as 48%. What surprised him even more was that this small plug-in even attracted the attention and investment of Y Combinator, laying the groundwork for the establishment of Jasper in the future.

YC is a well-known startup incubator, and joining the "YC Club" is one of the highest honors for startups. Dave, who suddenly reached the pinnacle of his life, founded a brand new software company Proof, rolling up his sleeves and preparing to change the world.

Two years later, the company laid off half of its staff, and Dave returned to his old business of paying for knowledge—everything he earned with shit luck was lost by Dave with his strength.

In 2021, Dave felt that paying for knowledge could no longer be done—the students in the class seemed to lack some talent, and the marketing copy they wrote always lacked a little "soul", which would inevitably damage their reputation in the long run. At this moment, Dave's "gears of fate began to turn" and hit the wind again:

In the "YC Club" group, the news of the internal test of GPT-3 was announced.

Altman, the founder of OpenAI, was the former CEO of YC, so he left the early internal test qualifications to YC start-ups; although Dave's software company failed, YC did not kick him out of the Slack group. **It turns out that luck does sometimes outweigh hard work. **

Dave thought that perhaps AI could be used to write marketing copy for people, so he and the two founders packaged GPT-3 into an MVP (Minimum Viable Product) and sent it to 10 regular customers by email.

Soon, he received feedback with only a few words, "I seem to see God" and "This is simply the next iPhone". The trainees waved banknotes one by one, begging Dave to continue to improve the product.

At this moment, the United States has one less knowledge payment company and one more AIGC unicorn.

High-rise building, banquet guests, building collapse

Although the seed users have blown Jasper AI into the sky, this is actually due to the credit of GPT-3. At that time, although GPT-3 was revolutionary, it was not easy to use and required extremely high-precision prompt words to generate high-quality copywriting. Jasper's job is to package GPT-3 into an easy-to-use product.

A comic that satirizes GPT's "shell"

As the GPT-3 API is gradually opened to more companies, similar companies have begun to emerge; Jasper AI can stand out from them because it has the right time, place and people. The "right time and place" is actually easy to understand: because Dave is a member of the "YC Club", he took the lead in obtaining the qualification for the internal test of GPT-3, and has a certain first-run bonus.

**The "human harmony" factor is actually more critical, referring to the support of a strong community. I have to say that Dave does have the temperament of a successful mentor. **

After teaching marketing for a few years, Dave turned many of his students into fans and created a community. This group of users is quite active, and will even take the initiative to help Dave decide company affairs. For example, after the random marketing plug-in became popular, Dave didn't want to make a fuss about it, claiming that unless the students were willing to pay $1,000 a year for it, dozens of people paid for it that day.

Now after Dave turned to AI, this group of iron fans took the initiative to act as "product managers".

As we all know, large language models such as GPT-3 have certain versatility, and developers can fine-tune them on this basis to adapt to more specific scenarios. It happened that Dave was born as a "marketing mentor" and had a lot of professional data in his hands; the Jasper team also absorbed a lot of feedback from hardcore fans, quickly iterated, and gradually differentiated products.

Today, Jasper is quite strong in the vertical scenario of "advertising and marketing", and has lowered the threshold of use through the design of "template".

Each segmented demand, such as Facebook marketing copy, Amazon product introduction, etc., has a corresponding template; users can generate ideal marketing copy by inputting prompt words according to the template. The template even includes options to define the tone of your copy, from humorous to elegant.

Jasper AI product interface

And this group of core users is not only good at speaking, but also unambiguous in paying; years of experience in selling courses has also allowed Dave to accumulate a fairly large customer list, knowing who to sell products to, so Jasper AI is very Soon the market opened up.

By the end of 2022, the number of Jasper AI customers will exceed 100,000, at least 3/4 of whom will spend more than $80 per month.

While Jasper AI has a super high payment rate, the cost can be negligibly low, which can be called a huge profit. At that time, the basic price of Jasper AI was $29/month, which could generate up to 20,000 words, and OpenAI only took less than $2 from it.

In October, Jasper AI received 125 million US dollars in financing, and the company scale has soared from 9 people at the beginning of the year to more than 160 people.

Not long after getting financing, Dave was invited to participate in a talk show. Faced with Jasper AI, which is rising at the speed of light, the host of the talk show expressed "Risby" deeply, but there is still a rather confusing question: Where is the moat of this company? Here's Dave's take on it:

"I've spent a lot of time thinking about this over the past 18 months. Then I realized that it's not just the giants that are successful. There are a lot of small companies like us that know their customers, build a great team, have good culture, maybe a bit lucky, kept executing, rolled out a second, third, fourth product, and ended up being a great company that executed fast and had a deep understanding of customers [7] 。”

Therefore, at the end of the program, Dave confidently stated that if OpenAI launches a new generation of models in the future, Jasper will still be the fastest runner when the starting gun is fired.

But a few days after the interview ended, the starting gun sounded like a bolt from the blue: the referee was leaving the game. ChatGPT no longer relies on complex prompt words, and only needs simple dialogues to generate copywriting.

Dave called Altman, the president of OpenAI, almost immediately, "Listen, I need to know what you are all planning to do." Altman tried his best to appease the anxious old customer, vowing that ChatGPT would not It will always be free, and OpenAI does not want to compete with its partners. The sudden popularity is purely an accident.

But this accident is enough to blow up a field.

Dimensionality Reduction Strike

Snazzy is Jasper's competitor, and its founder Chris Frantz is extremely pessimistic about the emergence of ChatGPT [8] , "No one thought that ChatGPT would be so good, so fast, and so free. Now you can generate a blog directly through the conversation. The big model was not easy to use, and the value of start-ups is to make up for this deficiency. Suddenly disappeared."

Jasper couldn't stand it anymore either. In June, traffic to Jasper's website fell by 50% compared to the beginning of the year. In July, Dave announced the layoffs on LinkedIn.

Although the industry is not optimistic, but Jasper has not completely surrendered.

Layoffs do not mean a complete failure. In fact, Jasper AI is still maintaining rapid iterations, including adding chatGPT-like dialogue functions, using more open source models, etc., and its reputation is quite good. A number of YouTube bloggers have made a comparative evaluation of Jasper AI and ChatGPT, and the results prove that Jasper AI is indeed slightly ahead in the specific scenario of advertising marketing-but only a few.

Jasper AI after adding chat function

The well-known investment institution a16z once divided the AI industry into three layers in an article, the bottom layer is the computing power infrastructure, the middle layer is the basic large model, and the upper layer is the AI application.

They believe: ** For a long time, most of the value of the AI industry has flowed to the bottom-level "shovel sellers", and all players in the remaining links currently lack long-term competitiveness. **

For upper-level AI applications, the biggest crisis comes from the "encirclement and suppression" of competing products from large companies. Because many of the current AIGC capabilities can actually be directly integrated into the original mature business of large companies. With mature application scenarios and a huge stock user base, large companies can easily beat entrepreneurs on the same track.

In fact, many large companies do the same. A few days ago, Adobe announced the launch of the "Sensei GenAI" series of services for marketers, and Microsoft's Microsoft 365 copilot is already on the way. With the huge user base of these two companies, the impact on Jasper may far exceed that of ChatGPT.

At the moment when everything is full of unknowns, it seems that the only way for start-ups to survive is to embrace change and change quickly.

Half a year after the launch of ChatGPT, Dave talked about the changes in Jasper in the past few months in an interview. He said that since the detonation of ChatGPT, the entire team has been exhausted, and the world seems to have been accelerated. Where can I go soon?

For domestic AI startups, Jasper's revelation may be more profound.

First of all, the squeeze from big factories will be more intense than that from overseas. As we all know, the overseas "basic-application" software ecology is relatively healthy, and OpenAI has not invested too much resources to directly end up as terminal Internet products. But on the contrary, domestic Internet companies have always had the habit of being "athletes", and domestic entrepreneurs will face more severe competition.

Second, start-ups that do not have their own large-scale model capabilities may be constrained. The reason why Jasper can still maintain a certain advantage over ChatGPT in many vertical copywriting fields is that it can fine-tune GPT-3. Domestic AI application layer companies may eventually need to master large-scale model capabilities.

In the stage of rapid technological change, anything that was previously considered a moat—such as product experience, user base, customer relationship, etc.—may be punctured by disruptive technologies. Especially in the field of software and the Internet with very short chains, the scenario of "a plug-in destroying a company" will happen again and again.

From a certain point of view: if you invest in 100 AIGC start-ups, the 10-year return rate of the combination may not be comparable to directly buying Microsoft + Nvidia stocks. **

** Epilogue **

Although Dave and Altman, the president of OpenAI, do not have much intersection, the two actually have many similar characteristics: both grew up in the Midwest of the United States, both are serial entrepreneurs, both call themselves idealists, and both are obsessed with "Big Play".

As mentioned earlier, Dave’s original intention of choosing to start a business was simply that he didn’t want to be a laborer. But with the unexpected popularity of the marketing plug-in, Dave gradually changed his mind.

He began to actively apply to join YC, trying to establish a software company that can really affect the world. During those days, Dave often took a hot bath with the two co-founders and fantasized about how to "conquer the world."

So when Dave saw GPT-3, he immediately made the MVP.

Daily life of the Jasper team

It's never a bad thing to fantasize about changing the world and then act on it, although only a handful of people actually do.

The emergence of Jasper AI does have a lot of luck; but if you always choose to walk with your head down, even if the sky really starts to drop, you will obviously not be able to pick it up.

AGI is undoubtedly a super wave, but the wind and waves will continue to destroy ship after ship before reaching the new continent.