"What’s critical in each of these cases is that these are not explicitly productivity improvements to existing employees for existing workflows. Instead, they take markets that were constrained by all the effort, friction, and cost of hiring and working with people, and unlock them. Exactly the type of market that gets overlooked by an incumbent."
One concept I like is that while the raw capacity of something like an LLM is increasing continuously over time, there's a hard threshold at which it crosses from being [not at all useful] to [useful] for a given application. Until we get true human-level AI-generated audio, ElevenLabs is impossible...but the second we do, it's a 10x improvement. Feels like part of the reason it's harder to spot these opportunities in advance.
It's easy to see large tech shifts as sustaining or efficiency innovations in the moment, because those uses are immediately obvious. In ~2000, Larry Ellison thought Amazon was a nice proof of concept, but he was holding his breath for when Neiman Marcus finally made the shift online. AI is going to sweep the floor with business process automation, but to think that the only impact of infinite, almost-free text, code, audio and imagery is a better ratio on a balance sheet, will seem laughable in hindsight. AI is an empowering technology – regardless of what the New Church of Finance thinks. The internet-native business are, on balance, the best businesses in the history of mankind. I'd be surprized if the same didn't turn out to be true for AI-native ones.
Something that interests me with AI is how it can be used to optimize processes in niche verticals. Here's one example I've come across lately: Industrial Data Labs is building an AI Sales copilot that helps inside sales teams at industrial distributors operate more efficiently, respond to more customer inquiries, and generate more sales. And it's only possible due AI.
I think of this kind of use case as 'applied AI,' as opposed to 'foundational AI' that you see with OpenAI, Google, etc. I think that most of the value in foundational AI will accrue to incumbents, largely because it is, or will become, commoditized tech. But for niche verticals, applied AI looks to be very promising.
It does make you think.. what do you make of investment opportunities to create AI "shields" for the receiving end? Along with the opportunities you describe, output is going to increase by orders of magnitude, and frankly, probably with quite high signal-to-noise ratio.
It's easy to picture a world with crisp and concise inputs creating the most imaginary media living in this synthetic, ballooned world, then being reduced to something crisp and concise on the other end.
thanks for the question Pelle. What Adobe is doing is a great example of "shields" (if I understand your point correctly). Giving designers the super powers of AI in a way that they can control. I can imagine both working -- the high signal to noise work still being done by professional designers/artists.
How do you see their relevance in the near future, and what strategies could they employ for defense? For instance, how many of the current AI companies might be overtaken by a more general AI, similar to what happened with AskPDF and ChatGPT4?
1. The target customers are trying to get a job done, but because they lack the money or skill, a simple, inexpensive solution has been beyond reach.
2. These customers will compare the disruptive product to having nothing at all. As a result, they are delighted to buy it even though it may not be as good as other products available at high prices* to current users with deeper expertise in the original value network. The performance hurdle required to delight such new-market customers is quite modest.
3. The technology that enables the disruption might be quite sophisticated, but disruptors deploy it to make the purchase and use of the product simple, convenient, and foolproof. It is the "fullproofedness" that creates new growth by enabling people with less money and training to begin consuming.
4. The disruptive innovation creates a whole new value network. The new consumers typically purchase the product through new channels and use the product in new venues.
Glad to read more from you recently! Most people are concerned with the supply/output but not yet clear what happens on the demand/consumption side. Let's consider just one example for simplicity: YouTube.
Supply: Let's assume all these text, video, editing and image tools let current creators create more content, allow more people to create content, and each unit of content created is 'better'. Supply and quality of 'content' is gonna go up a LOT. This is a great thing overall...but....
Demand: On the human side, (mostly) humans consume the content and the rate of growth of humans will be more limited. Each human also has a limit on leisure / content hours, and I don't expect that to change much short-term.
How will such content output be consumed / discovered / curated? What new modalities or improvements will consumers need to be able to pick through the supply growth / or better process such volume? In this example, how would let's say YouTube as a product need to change to address that supply/demand imbalance? Yes, some better personalization, yes perhaps AI can summarize some of this net new content into the spark notes but overall something might have to change / give.
What needs to change on the consumption side to enable users to adapt / better absorb that content?
"What’s critical in each of these cases is that these are not explicitly productivity improvements to existing employees for existing workflows. Instead, they take markets that were constrained by all the effort, friction, and cost of hiring and working with people, and unlock them. Exactly the type of market that gets overlooked by an incumbent."
🎯
One concept I like is that while the raw capacity of something like an LLM is increasing continuously over time, there's a hard threshold at which it crosses from being [not at all useful] to [useful] for a given application. Until we get true human-level AI-generated audio, ElevenLabs is impossible...but the second we do, it's a 10x improvement. Feels like part of the reason it's harder to spot these opportunities in advance.
very insightful point, Ben. Also speaks to the opportunities here evolving continuously as the technology improves...
It's easy to see large tech shifts as sustaining or efficiency innovations in the moment, because those uses are immediately obvious. In ~2000, Larry Ellison thought Amazon was a nice proof of concept, but he was holding his breath for when Neiman Marcus finally made the shift online. AI is going to sweep the floor with business process automation, but to think that the only impact of infinite, almost-free text, code, audio and imagery is a better ratio on a balance sheet, will seem laughable in hindsight. AI is an empowering technology – regardless of what the New Church of Finance thinks. The internet-native business are, on balance, the best businesses in the history of mankind. I'd be surprized if the same didn't turn out to be true for AI-native ones.
love this. Thank you Max!! so well put.
Something that interests me with AI is how it can be used to optimize processes in niche verticals. Here's one example I've come across lately: Industrial Data Labs is building an AI Sales copilot that helps inside sales teams at industrial distributors operate more efficiently, respond to more customer inquiries, and generate more sales. And it's only possible due AI.
I think of this kind of use case as 'applied AI,' as opposed to 'foundational AI' that you see with OpenAI, Google, etc. I think that most of the value in foundational AI will accrue to incumbents, largely because it is, or will become, commoditized tech. But for niche verticals, applied AI looks to be very promising.
Hi Sarah, great writing as always.
It does make you think.. what do you make of investment opportunities to create AI "shields" for the receiving end? Along with the opportunities you describe, output is going to increase by orders of magnitude, and frankly, probably with quite high signal-to-noise ratio.
It's easy to picture a world with crisp and concise inputs creating the most imaginary media living in this synthetic, ballooned world, then being reduced to something crisp and concise on the other end.
thanks for the question Pelle. What Adobe is doing is a great example of "shields" (if I understand your point correctly). Giving designers the super powers of AI in a way that they can control. I can imagine both working -- the high signal to noise work still being done by professional designers/artists.
How do you see their relevance in the near future, and what strategies could they employ for defense? For instance, how many of the current AI companies might be overtaken by a more general AI, similar to what happened with AskPDF and ChatGPT4?
Classic new market disruption
1. The target customers are trying to get a job done, but because they lack the money or skill, a simple, inexpensive solution has been beyond reach.
2. These customers will compare the disruptive product to having nothing at all. As a result, they are delighted to buy it even though it may not be as good as other products available at high prices* to current users with deeper expertise in the original value network. The performance hurdle required to delight such new-market customers is quite modest.
3. The technology that enables the disruption might be quite sophisticated, but disruptors deploy it to make the purchase and use of the product simple, convenient, and foolproof. It is the "fullproofedness" that creates new growth by enabling people with less money and training to begin consuming.
4. The disruptive innovation creates a whole new value network. The new consumers typically purchase the product through new channels and use the product in new venues.
Glad to read more from you recently! Most people are concerned with the supply/output but not yet clear what happens on the demand/consumption side. Let's consider just one example for simplicity: YouTube.
Supply: Let's assume all these text, video, editing and image tools let current creators create more content, allow more people to create content, and each unit of content created is 'better'. Supply and quality of 'content' is gonna go up a LOT. This is a great thing overall...but....
Demand: On the human side, (mostly) humans consume the content and the rate of growth of humans will be more limited. Each human also has a limit on leisure / content hours, and I don't expect that to change much short-term.
How will such content output be consumed / discovered / curated? What new modalities or improvements will consumers need to be able to pick through the supply growth / or better process such volume? In this example, how would let's say YouTube as a product need to change to address that supply/demand imbalance? Yes, some better personalization, yes perhaps AI can summarize some of this net new content into the spark notes but overall something might have to change / give.
What needs to change on the consumption side to enable users to adapt / better absorb that content?