AI with unicorns
Image generated by Stability AI's DreamStudio Beta

The art of the DALL-E: A toy or the next wave of startup innovation?

Protocol Pipeline

Hello, and welcome to Pipeline. I’m Biz Carson, and I’m celebrating my 5-year wedding anniversary this weekend. Thank you to my husband David for also sacrificing your Friday nights and Saturday mornings for the last few years to allow me to get this newsletter out the door.

This week in the startup world: it’s Adobe versus Canva now, more CEOs step down, and a look at the investor interest in generative AI imagery.

A DALL-E dalliance

It’s a famous startup saying that the next big thing will start out looking like a toy. And there’s no toy that VCs have been more excited about playing around with recently than DALL-E and other generative AI image tools.

Put in a few key words into a tool like Midjourney, Stable Diffusion, or DALL-E and it’s easy to see why the whimsical (and often wacky) images have captured investors’ imagination. An AI-generated artwork even recently won an art competition at the Colorado State Fair, a result that didn’t go over well among more traditional artists. It’s become disruptive enough that this week Getty announced a ban of AI-generated images on its platform, following similar moves by some online art communities.

What looks like an interesting art tool has become a prime feeding ground for investors. Investor interest has been nearly overwhelming for Poly’s Abhay Agarwal, who is building a “DALL-E for design assets” company. “It has literally been like dropping yourself into the Ganga River and fully being bathed in it,” Agarwal said of the interest. He’s already had over 80 meetings with VCs and is only halfway done following YC’s Demo Day.

  • The challenge now for investors is finding the business case in AI-generated imagery. Already, some companies like Stitch Fix have been experimenting with the technology, but with mixed success. “I feel quite strongly that these technologies are quite world-changing,” Khosla Ventures partner Kanu Gulati told me. “They’re still early. A lot of their shortcomings are known, but the community is super, super active and trying to resolve them.”
  • Perhaps unsurprisingly, the initial startup applications have been around design, marketing and ecommerce, like a company doing AI-generated stock imagery or a startup building AI models for fashion brands so they can skip photoshoots. Gulati has invested in startups like Rosebud, which is doing AI-generated photos and videos (including NFTs), while Khosla Ventures has directly backed research lab OpenAI, the creator of DALL-E. Poly is pitching itself as a way for designers to use AI to generate textures.
  • Already looking ahead, Gulati thinks AI imagery will be used in combination with other forms of generative AI like text, and that’s where more value can be created. “There will be huge industries out there giving Adobe a run for their money because of using these latest technologies,” Gulati said. “And these will be built on a new stack of AI-first companies.”

The hype wave is similar to GPT-3, a generative AI text tool with an API that businesses can build off of. The problem is that investors can easily fall into the trap of thinking the two generative models are the same.

  • For generative text, there can be a lower bar for quality and also a lower bar for utility. If the AI makes mistakes, it’s easy to clean up typos. But plenty of people can also write their own mediocre copy if needed, so the value of some tools is diminished if replacing a human with AI doesn’t really save much cost or work.
  • The bar for images is much higher, because if an image comes back where something is wrong (like the unicorn with seven horns pictured in this article) then it has to be tossed out — you can’t correct it easily. But at the same time, the utility is high because, frankly, most humans can’t create a drawing anywhere near the quality of the output of the AI, Agarwal explained.
  • “For text modeling, someone can do a mediocre job of it on their own,” Agarwal said. “With image modeling, you can't. 99.9% of people in this world cannot create a convincing illustration, even given an infinite amount of time.”

Just because it’s magical doesn’t mean it can magic away its shortcomings. As Charlie Warzel pointed out in a smart piece, “What feels like magic is actually incredibly complicated and ethically fraught.”

  • The black box algorithms behind much of the programs have already raised serious concerns about copyright and other legal claims as it’s not known what imagery the models were trained off of. Stable Diffusion recently did release its training model, and much of it came from Pinterest imagery and Thomas Kinkade’s art, per Andy Baio’s analysis.
  • Already, there’s a lot of bias in the models. Run a search for a startup founder or venture capitalist and it almost always returns a white male as the image. Even a search that included “teacher,” a predominantly female profession, returned images of men. “Bias will continue to be a big challenge, which investors and founders have to solve before these become sustainable enterprises,” Gulati said.
  • And with every tool on the internet, what can be used for good can also be used for evil. Stable Diffusion recently open-sourced its technology in a way that could allow people to circumvent safeguards and create pornography, deepfakes, and violent imagery — something tools like DALL-E block. There are websites and Discord forums popping up specifically around AI-generated pornography already, and people posting images of Bernie Sanders in a “Mad Max” deepfake.
Creating a future for generative AI startups won’t be as easy as painting a picture of the opportunity. Founders and investors will have to both take responsibility for understanding the shortcomings of generative AI and solving them. It takes more than “hustling and flipping when you see a quick opportunity to leverage an open-source technology,” said Agarwal. Instead, he argued technologists need to become stewards of the technology and build it for whatever business application is needed. For Poly, that means creating and training its models around textures and design elements so that it can responsibly tailor the model in a way that allows it to build a business. “I don't believe that once a model was released into the open-source public that somehow that means that everybody can jump on that and start using it for whatever use case,” Agarwal said.


“The biggest secret to success in VC is to get lucky in your first five years,” tweeted Floodgate’s Mike Maples. It turns out he’s not wrong, at least according to this 2019 Harvard paper that showed early IPOs led to higher rates of IPOs and better deal flow in the future.

R.I.P. Robot Scoble. Speaking of generative AI, a Twitter bot based on Robert Scoble’s tweets briefly popped up this week to talk about how Apple was going to crush the world. Sadly the Robot Scoble bot lasted only a few days before the internet abused it and it “passed away in a tragic Metaverse car accident.”

“I realized it was going to take years to break even,” said one Helium customer after shelling out $500 for a hotspot. A Forbes investigation into the Web3 startup Helium shows it was mostly insiders, not real public customers, who amassed a fortune — begging the question of whether the startup is just a lot of hot air.

“I feel compassion for Ms. Holmes and Mr. Balwani, and even more so for the members of their families who were not responsible for their conduct but will be affected by the punishment they may receive,” Adam Rosendorff, a key witness for the prosecution in the Theranos case, said in a newly filed affidavit. In a bout of witness remorse, Rosendorff said everything he said on the stand was true, but clearly the case was weighing on him as he visited Elizabeth Holmes’ house. Holmes is using Rosendorff’s unusual visit in one of her bids to get a new trial ahead of her sentencing on Oct. 17.


USD Coin (USDC) is the institutional grade stablecoin. Monthly attestations show exactly what reserves back USDC, and businesses all over the world are using USDC to build the next generation of financial services and global payment applications.

Learn why institutions trust USDC at Circle’s Transparency & Stability Hub

Join us

The global innovation race is well underway. What is the U.S. administration doing to stay ahead, and where is it falling short? What is the status of funding by Congress and in statehouses, and which areas still need investment? Is the U.S. doing enough to attract and retain top tech talent from around the world? And let’s not forget about China. Join the Protocol Policy team on Sept. 27 as it dives into the U.S.’s national strategy on innovation, what’s working, what isn’t, and what policy changes we can expect from the year ahead. RSVP here.

Inside track

“What Pinterest was to images, TikTok is becoming for short-form video,” said a16z’s Connie Chan. While most people (including myself) use TikTok to scroll through hours of videos on end, it could follow in the footsteps of its Chinese sister app Douyin by turning its videos into searchable and buyable experiences.

When NFX’s Pete Flint started Trulia in 2005, it was still the Real Estate 1.0 era. Now, after a wave of startups that changed real estate transactions, he believes we are entering a third era of real estate: the ownership revolution.

Everyone hates that recurring meeting that no longer serves a good purpose, but the inertia of an organization means it's often left on people’s calendars. That’s why ex-Facebooker Molly Graham argues that companies need to be as good at subtraction as they are at addition. Try starting with a complete calendar wipe to see how much time you regain.

When a funding environment gets tough, you start hearing words like “cram-down” or a “dirty” term sheet come around. To better understand all the deal sweeteners that can be added in, a16z’s Shangda Xu, Alex Immerman and David George wrote a guide to fundraising terms that pop up in a tough environment, along with some hypothetical scenarios on how they might affect a company’s outcome.

Need to know

Venture capitalists are sitting on a ton of cash. With $290 billion in dry powder, Jon Sakoda’s analysis showed that this year is likely to be a moderate slowdown, but 2023 and 2024 will still be record years for funding.

Chamath Palihapitiya shut down two SPACs. The king of SPACs couldn’t find a target for the vehicles, joining a list of names of those — like Bill Ackman — who have closed down the acquisition companies in a dry year for the markets.

I guess we don’t get flying cars after all. Larry Page’s flying car startup Kitty Hawk announced it was winding down this week. Its joint venture with Boeing, Wisk, will continue, but it’s locked in a bitter lawsuit with rival Archer.

Inside the end of Pollen. The events startup had raised $200 million but let its staff go unpaid by the end in a dramatic collapse of the company.

Moves: Kraken’s CEO Jesse Powell is not “taking a jet ski” but is stepping back from his role as CEO after growing tired of dealing with the bureaucracy of running a finance business. “I'm much more like a zero-to-one kind of guy,” he told Protocol. Bird CEO Travis VanderZanden also stepped down this week as the scooter startup faces a dismal outlook. Klarna did a second round of layoffs. Beyond Meat’s COO was arrested after biting someone’s nose.

From Protocol: After Figma’s sellout, it’s Adobe versus Canva for dominance in design software.

Also from Protocol: It may be a crypto winter, but there’s still a lot of investor demand for crypto infrastructure providers.

Also, also from Protocol: Ripple’s co-founder funded a 1,000-camera surveillance network in SF. Now the city’s police department can watch it too.

Your weekend reading: Coinbase launched its nonprofit with great fanfare, promising to use ambassadors to give away crypto in developing nations to promote cryptocurrencies around the globe. As Fortune found, the charity GiveCrypto came to “represent the dark side of crypto-based charity — a mess of illogical coordination, misplaced assurances, and unpaid labor.” Or as one professor put it, the worst bits of the gig economy and cryptocurrency put together for a “horrific” outcome.


USD Coin (USDC) is the institutional grade stablecoin. Monthly attestations show exactly what reserves back USDC, and businesses all over the world are using USDC to build the next generation of financial services and global payment applications.

Learn why institutions trust USDC at Circle’s Transparency & Stability Hub

Thanks for reading! If you like what you’re reading, sign up here to get it in your inbox. Send story tips and newsletter feedback to

Recent Issues

VCs rethink the office