Drawing on Insight Partners’ experience working with AI startups since 2014 — with $4 billion invested in the sector so far — Managing Director Lonne Jaffe discusses artificial intelligence-related lessons learned, and best practices brought to life, including the importance of explainability and the promise of next-gen MLOps stacks.
These insights came from our ScaleUp: AI event in October 2023, an industry-leading global conference that features topics across technologies and industries. Watch the full session below:
From a tool for prediction to a tool for creation
The shift in AI development, from tools that predict and classify to systems that create in meaningful and impactful ways, is upon us. Initially, AI was used for use cases like fraud detection and image classification, but recent developments have enabled machines to write software, draft blog posts, and generate images.
“In the old days, humans would show up at work and type on keyboards with their fingers, and that’s how we got the software. And then starting a couple decades ago… data would go into the machine and software would come out,” Jaffe explains.
The latest AI systems that can write a poem or generate images are now systems of creation, distinguishing them from the prior era of prediction and classification systems.
“We can now make decisions, take actions, and build something new from scratch, which was not possible with the previous generation of technology.”
AI-driven price decreases may disrupt industry pricing but also expand markets
The advent of AI has significantly decreased the price of numerous services in the economy. This could lead to disruption in certain industries, but also to potential market expansion in sectors with high elasticity of demand. A key consideration for companies is how to capture the value generated by these price decreases.
“One way to think about how this is… it’s significantly decreasing the price of a huge number of things that go on in the economy. The cost and the price — and it’s not the first time that’s happened.”
The technology industry is used to relentless price declines, and when prices go down, some things become significantly more valuable, which economists call economic complements.
AI transformation has favored incumbents over startups
“There’s a knob, and the knob is typically set to startup for most big technology shifts. It’s a balance-of-power knob, and that knob has been turned toward the incumbents significantly,” Jaffe says. The ability of existing companies to integrate AI into their operations has shifted the balance of power in their favor, as opposed to startups. This shift is due to incumbent firms’ ability to leverage existing customer bases, data assets, and reputations to incorporate AI into their products and services.
Jaffe further explained that by incumbents, he was referring to existing companies, which may already enjoy some basic platform of scale effects, strong talent, and the ability to build a reputation for reliable, beautifully crafted products.
Significant challenges also present opportunities
Despite the rapid advancements in AI, there are still significant challenges that need to be addressed. These include hallucination (AI generating false or misleading information), alignment issues, potential cyber threats, and unresolved legal issues regarding IP and copyright. However, these challenges can also present opportunities for startups to provide solutions.
“There are a lot of really significant challenges around generative AI that we’re seeing. A couple of points I’ll make about this though… challenges often represent startup opportunities.”
Some of these challenges have proven less enduring than expected, with significant mitigations achieved within the year, such as privacy concerns and the inability of AI to do math. “Generative AI has allowed us to think of potential solutions in diverse areas such as healthcare, transportation, and energy. Startups can use this to their advantage, by developing innovative solutions which can make a real difference in the world. The opportunities for startups are immense, and the potential rewards are as well,” Jaffe says.
Startups and ScaleUps have agility on their side
Jaffe emphasized the ease with which AI can be integrated into businesses, which has led to the rapid development and monetization of AI products. This has been particularly clear in existing companies that have been able to adapt and leverage AI technology. However, startups are in a particularly advantageous position.
Because they are not constrained by existing infrastructure and processes, startups and ScaleUps can innovate quickly and efficiently, creating products or services that leverage AI in unique ways. This has enabled startups to outpace existing companies in creating AI-based solutions that have a real impact. Ultimately, AI provides a great opportunity for startups to build products that are both innovative and profitable.
“There’s one class of startup that can be kind of interesting…Kevin Scott from Microsoft has coined a phrase for a certain type of startup, which he refers to as ‘making the impossible possible, but still difficult.’ This is different from the generative AI systems which strive to make the hard easy.” Jaffe says. For example, Profluent uses chat GPT-like language model capabilities for protein design, a task that was previously impossible. This type of innovation is particularly exciting because it allows startups to solve problems that were previously unsolvable. By leveraging AI, startups can create solutions that not only address existing problems but also create entirely new solutions that weren’t possible before.
Insight has invested in Profluent.