Featured Article

So you want to launch an AI startup

It’s the best of times, it’s the worst of times

Comment

Illustration of a man holding an cog with a light bulb and a woman fitting her cog together with it with a dollar sign.
Image Credits: sorbetto / Getty Images

It seems like it’s the best of times for founders thinking about launching an AI startup, especially with OpenAI releasing ChatGPT to the masses, as it has the potential to really put AI front and center in business and perhaps everything we do technologically. Who wouldn’t want to launch a startup right now with the energy and hype surrounding the industry?

But it also could be the worst of times for founders thinking about launching an AI startup, especially one that can grow and be defensible against incumbents in a fast-changing environment. And that’s a real problem for companies thinking about this area: AI is evolving so rapidly that your idea could be obsolete before it’s even off the ground.

How do you come up with a startup idea that can endure in such a challenging and rapidly evolving landscape? The bottom line is that the same principles that apply to previously successful startups apply here, too. It just may be a bit harder this time because of how quickly everything is moving.

A bunch of successful founders and entrepreneurs spoke last week at the Imagination in Action conference at MIT. Their advice could help founders understand what they need to do to be successful and take advantage of this technological leap.

What’s working?

CB Insights compiled data from 2021 and 2022 to understand where VC investment money has been going when it comes to generative AI startups. Given the recent hype around this area, it’s reasonable to think that the volume of investment will increase, and perhaps the allocation will be different, but this is what we have for now.

CB Insights chart on where generative AI investment went in 2021 and 2022.
Image Credits: CB Insights

But will these numbers still be relevant for 2023 and beyond — and how do you build a successful AI startup that attracts investment dollars?

Sam Altman, OpenAI co-founder and CEO, who spoke via Zoom at the conference, said that you have to think about this next generation of startups a little differently, but ultimately it comes back to the basics of building any successful startup.

“I think you never want to lose sight of vision and focus on the long term. But the very tight feedback loop of paying attention to what is working and what is not working, and doing more of the stuff that’s working and less of the stuff that’s not working, and just very, very careful user observation can go super far,” he told conference attendees.

His advice is to just listen to your customers and respond to their needs. “I can speculate on ideas, you can speculate on ideas. None of that will be as valuable as putting something out there and really deeply understanding what’s happening and being responsive to it,” Altman said.

Vinod Khosla, the longtime venture capitalist, founder at Khosla Ventures and early OpenAI investor, speaking via Zoom at the event, said that speed of change is a huge factor in building an AI startup right now: Your idea could be obsolete before it’s even off the ground, which means, you have to think about ideas that can stand the test of time.

“I looked at the recent YC batch, and when OpenAI released one feature, which is plug-ins, around half the batch got obsoleted. Literally, they were doing something minor with ChatGPT or GPT-4,” Khosla said. Working backward from ChatGPT, or whatever the flavor of the month happens to be, won’t necessarily work for AI startups because your idea won’t be built to stand up to changes coming from OpenAI itself, and presumably other larger incumbents, he said.

“So there are ways to build differentiation both in the application domain and in the fundamental model domain – and then assembling the right teams to go after these products,” Khosla said.

Speaking on a panel made up of successful founders, Deep-Mind CEO and co-founder Dileep George said that you have to build something that stands up to deployment outside of the lab. “What you really need to be worried about is the demo-to-deployment gap that all AI systems have. It’s very easy to create a demo in the lab, but to make it a deployment to get customer value, there’s a big gap in real-world deployment. And that’s where all the pain in use cases that come up in AI, and you have to be aware of that and mitigate that from day one,” George said.

Stick to basics

Steve Papa, who founded Endeca, a database company that was sold to Oracle in 2011, and later founded Toast, the restaurant management tool, says that regardless of the technology, the same success principles always apply and the same types of startups tend to be successful over time. The trick here is to figure out how AI is going to make existing tooling better.

“We’re old enough to have been able to see in the ’90s, the aughts and the teens, and the same [types of] companies are rebuilt, they get acquired, and then a whole new generation [comes along to solve] each of these classes of problems. And there’s got to be one for every one of those where AI is gonna fuel it, accelerate it, make it more capital efficient, make it go faster,” Papa said.

And sometimes you just need to look at ways to plug in AI to make existing enterprise software better, faster and easier to use, said Brian Halligan, who co-founded Hubspot and is now working on a new venture firm called Propeller to invest in green tech related to the world’s oceans.

Generative AI could transform the way we interact with enterprise software

“Every piece of enterprise software is going to go through [a] transition similar to how the world of DOS went to Windows [to change] the way you interact with the software,” he said.

But you also need to prepare by building resilience and defensibility into your startup. Khosla said that if you are going to build a successful company, you have to do it by thinking about what’s coming next to large language models, or even maybe something that doesn’t involve large language models, but another approach to building AI. Regardless, you have to think creatively, and look beyond whatever is happening right now.

“I suspect there are some people in the audience who say, this is not the end all and large language models are not the only way to do AI. So we are actively investing in alternatives approaches to large language models and bigger and bigger models,” he said.

If you look at the growth in capability we have seen from GPT-2 to GPT-3 to GPT-4, it’s likely only to increase as we get to the next versions of the model, he said. Smart founders will be thinking about future capabilities instead of building for current ones.

“For the entrepreneurs in the audience, I’d say you need to skate where the puck is going, not where it is,” he said.

More TechCrunch

Cargo ships docking at a commercial port incur costs called “disbursements” and “port call expenses.” This might be port dues, towage, and pilotage fees. It’s a complex patchwork and all…

Shipping logistics startup Harbor Lab raises $16M Series A led by Atomico

AWS has confirmed its European “sovereign cloud” will go live by the end of 2025, enabling greater data residency for the region.

AWS confirms will launch European ‘sovereign cloud’ in Germany by 2025, plans €7.8B investment over 15 years

Go Digit, an Indian insurance startup, has raised $141 million from investors including Goldman Sachs, ADIA, and Morgan Stanley as part of its IPO.

Indian insurance startup Go Digit raises $141M from anchor investors ahead of IPO

Peakbridge intends to invest in between 16 and 20 companies, investing around $10 million in each company. It has made eight investments so far.

Food VC Peakbridge has new $187M fund to transform future of food, like lab-made cocoa

For over six decades, the nonprofit has been active in the financial services sector.

Accion’s new $152.5M fund will back financial institutions serving small businesses globally

Meta’s newest social network, Threads, is starting its own fact-checking program after piggybacking on Instagram and Facebook’s network for a few months.

Threads finally starts its own fact-checking program

Looking Glass makes trippy-looking mixed-reality screens that make things look 3D without the need of special glasses. Today, it launches a pair of new displays, including a 16-inch mode that…

Looking Glass launches new 3D displays

Replacing Sutskever is Jakub Pachocki, OpenAI’s director of research.

Ilya Sutskever, OpenAI co-founder and longtime chief scientist, departs

Intuitive Machines made history when it became the first private company to land a spacecraft on the moon, so it makes sense to adapt that tech for Mars.

Intuitive Machines wants to help NASA return samples from Mars

As Google revamps itself for the AI era, offering AI overviews within its search results, the company is introducing a new way to filter for just text-based links. With the…

Google adds ‘Web’ search filter for showing old-school text links as AI rolls out

Blue Origin’s New Shepard rocket will take a crew to suborbital space for the first time in nearly two years later this month, the company announced on Tuesday.  The NS-25…

Blue Origin to resume crewed New Shepard launches on May 19

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

In the coming months, Google says it will open up the Gemini Nano model to more developers.

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people