Dennis Gretch
Former CTO of Plaid and a serial entrepreneur who founded Town, a company dedicated to applying AI to email and calendar management.
Timeline (last 90 days)
Dennis Grelz is the former CTO of the fintech company Plaid.
Grelz believes that the ceiling of the tax filing business is limited by the emergence of large models, and that incumbents can use the same set of models to automate existing clients' filings and grab share faster than startups.
After the end of the first tax season, Grelz and his co-founder decided to shut down the tax team and switch to a market more worth pursuing.
Grelz invokes the Jevons paradox to explain that as large-model token prices fall, people will find more and more scenarios for using AI at work, and total token consumption and total AI value will expand accordingly.
Grelz's judgment: if the customer is not the user, you are likely to build a bad product that makes decision-makers happy but users miserable, so the focus should be on making the product good enough, entering the mid-to-low-end market through self-serve growth, and then winning enterprises from a security and compliance angle.
Grelz believes the agent category already has millions of users worldwide but is still at a very early stage of adoption, and it may take 6 to 18 months for most people to first experience the magic of AI themselves before considering internal corporate deployment, and enterprise deployment is more complex than the iPhone because agents need access to data and the APIs that support the business.
All Facts (6)
Dennis Grelz is the former CTO of the fintech company Plaid.
50%UnverifiedGrelz believes that the ceiling of the tax filing business is limited by the emergence of large models, and that incumbents can use the same set of models to automate existing clients' filings and grab share faster than startups.
50%UnverifiedGrelz invokes the Jevons paradox to explain that as large-model token prices fall, people will find more and more scenarios for using AI at work, and total token consumption and total AI value will expand accordingly.
50%UnverifiedGrelz's judgment: if the customer is not the user, you are likely to build a bad product that makes decision-makers happy but users miserable, so the focus should be on making the product good enough, entering the mid-to-low-end market through self-serve growth, and then winning enterprises from a security and compliance angle.
50%UnverifiedAfter the end of the first tax season, Grelz and his co-founder decided to shut down the tax team and switch to a market more worth pursuing.
50%UnverifiedGrelz believes the agent category already has millions of users worldwide but is still at a very early stage of adoption, and it may take 6 to 18 months for most people to first experience the magic of AI themselves before considering internal corporate deployment, and enterprise deployment is more complex than the iPhone because agents need access to data and the APIs that support the business.
50%