June 8, 2023
Cautious optimism around generative AI
Like everyone else in the world QED’s conversations have been awash in generative AI discourse. Unlike most of our peers, we’re resisting jumping to conclusions.
It’s been obvious to us that machine learning and AI would be a massive force in our economy – the use cases were too many, the potential pathways for value were too obvious and our portfolio companies were constantly finding ways to use machine learning widgets to increase value – whether that was matching of business addresses in anti-money laundering (a use case from 2018) to data extraction and small business lending fraud (Ocrolus) or transaction labeling at scale (Ntropy).
Moreover, ChatGPT, Bard and its competitors are breakthroughs for user experience and accessibility. What’s less obvious to us is that this particular moment with generative AI and large language models is a transformative moment for financial services.
Here’s how we’re thinking about navigating this territory:
1. Generative AI will be a massive accelerant for coding, marketing and other content. We are telling all of our companies to begin experimenting.
2. Particularly for customer service, we think that opportunities are enormous but financial services requires accuracy and compliance – something LLMs are demonstrably bad at.
One of our portfolio companies, Coru, has already built and is starting to scale a real-time coaching service that has solved the problem of using Chat GPT, while controlling its output for financial advice, customer service, sales or any other type of assistance the user may need.
3. We still don’t know how quickly (or if) the generalist AI-industry leaders will be able to conquer domain-specific use cases. In this context, the companies who may benefit most from generative AI may also be those who are most at risk.
4. We’re not allocating specific investment dollars to an AI mandate, at the margin we’re skeptical that the new wave of AI-hype companies will be backable. Instead we’re more interested in how companies are using this new tool – to accelerate or to unlock previously hard problems.
5. More generally, we are confident that generative AI will upset the value chain, so we’re even more focused on moats that we think will persist – access to scarce resources, network effects, deep integrations and sticky contracts.
As always, our core focus is to keep learning. To that end, we’ve written a very simple survey on the strategic questions for generative AI in fintech. We’d love your input!
By QED Partner Amias Gerety and QED Principal Adams Conrad.
Like everyone else in the world QED’s conversations have been awash in generative AI discourse. Unlike most of our peers, we’re resisting jumping to conclusions.
It’s been obvious to us that machine learning and AI would be a massive force in our economy – the use cases were too many, the potential pathways for value were too obvious and our portfolio companies were constantly finding ways to use machine learning widgets to increase value – whether that was matching of business addresses in anti-money laundering (a use case from 2018) to data extraction and small business lending fraud (Ocrolus) or transaction labeling at scale (Ntropy).
Moreover, ChatGPT, Bard and its competitors are breakthroughs for user experience and accessibility. What’s less obvious to us is that this particular moment with generative AI and large language models is a transformative moment for financial services.
Here’s how we’re thinking about navigating this territory:
1. Generative AI will be a massive accelerant for coding, marketing and other content. We are telling all of our companies to begin experimenting.
2. Particularly for customer service, we think that opportunities are enormous but financial services requires accuracy and compliance – something LLMs are demonstrably bad at.
One of our portfolio companies, Coru, has already built and is starting to scale a real-time coaching service that has solved the problem of using Chat GPT, while controlling its output for financial advice, customer service, sales or any other type of assistance the user may need.
3. We still don’t know how quickly (or if) the generalist AI-industry leaders will be able to conquer domain-specific use cases. In this context, the companies who may benefit most from generative AI may also be those who are most at risk.
4. We’re not allocating specific investment dollars to an AI mandate, at the margin we’re skeptical that the new wave of AI-hype companies will be backable. Instead we’re more interested in how companies are using this new tool – to accelerate or to unlock previously hard problems.
5. More generally, we are confident that generative AI will upset the value chain, so we’re even more focused on moats that we think will persist – access to scarce resources, network effects, deep integrations and sticky contracts.
As always, our core focus is to keep learning. To that end, we’ve written a very simple survey on the strategic questions for generative AI in fintech. We’d love your input!
By QED Partner Amias Gerety and QED Principal Adams Conrad.