Evaluating New Technology

"When someone comes up to me and says, 'We have this one click agent, it's going to be deployed in your system.' ... I would almost be skeptical because it's just not possible. And that's not because the models aren't there, but because enterprise data and infrastructure is very messy... I would rather go with a company that says, 'We're going to build this pipeline for you,' and that will learn over time."

"The models couldn't do some things one year ago. I mean, image generation was full of spellings or reasoning. You just couldn't have deeper and smarter answers. You couldn't do data analysis. So my impression of it from change, trying it a few months ago, that prior needs to be updated. And it's hard to do that, right? You have to do something almost counterintuitive and against the grain to say, 'No, no, ignore what you learned about what this can or cannot do.' The baby just grew up to be a 15-year-old in a month."

"you really need to bet on a platform or some app server type layer that allows you to swap things in and out and not really be beholden to anything, any one technology or any one tool because the reality is the whole thing is going to change."

"I have this adage I always say, which is tools are just meant to solve problems. And the problem set for marketing technologists and business technologists is you focus on the tools."

"It's B and B as opposed to BVB. So, build and buy as opposed to build versus buy. People all the time just think the second that you're talking about implementing a tool or procuring a solution, it's, Hey, I want to build this thing or I want to buy this really expensive thing. Build versus buy is a very narrowly constricting decision tree. If it's only build versus buy, then you've already made the decision that you can only do one or the other... Build and buy means that both of you can win"

"if you're not technical, really lean into it and build something simple, learn how to build something simple for yourself, demystify the technology. That experience will take you far. I love telling people that literally don't even know what HTML is... you could build a clone of Twitter using a tutorial on Rails"

"GraphQL... is one of the things that is both popular and thought relatively poorly of by most of the senior people that I know... if you're seriously thinking about it and you're not Facebook, you may really want to make sure you know what problem you're trying to solve because the impression that I have... is that GraphQL is kind of trying to promise front-end engineers that they don't really have to collaborate with backend engineers."

"I do still think studying computer science is a different answer than learning to code... computer science is a wonderful major to learn systems thinking."

"We had someone at one point handle whatever tools connected to Pendo and make sure that those systems are set up for maximum outcomes for the product manager. So, Pendo's connected to Salesforce. We're connected to Looker. We're connected to all these different."

"My first thing that I open is o3. I'm a ChatGPT boy... it has memory. And I just love that... I think Claude Opus is... Claude Code, everyone inside Every, that's basically what we use... Gemini... It's incredibly powerful and it's incredibly cheap, which is great."

"The savings and costs that there might be in replacing a vendor tool by something you build in-house is probably not worth it in the mental bandwidth that you've lost and the amount of the team's technical focus that's being taken away. ... I would always come back to what is the reason we're doing this? Why does it matter to us and to our customers?"

"I would say really try and use these tools yourself. ... we learn a lot about how our own workflow can change, and that's going to tell you so much more about how are you going to change your organization's workflow than if you're reading a bunch of think pieces on LinkedIn."

"I try to use as many different AI products, including not Airtable, as I can... I try to invent little, almost like side projects of my own, to have a real reason to use these products."

"Before React wins, there's a competing technology called Web Component from Google... we're betting on that technology. And then we realize because it's so new, it's just so unstable. It don't know where the bug come from. It's from your source code or from the underlying libraries? Then we have to restart the company, rebuild the whole thing."

"One of the things that we used to talk about all the time at Google on Maps was how would we design a navigation product when people are in self-driving cars?... It's going to take years, but it is just such a different paradigm. It's like computers talking to computers, algorithms talking to other algorithms. Then there's a human in the mix in terms of serving the human at the end, but it's like the human is not initiating that much."

"The modern growth stack... is the evolution of what you do with the data. So these are the workflows that the data enables to drive the business forward for product growth and revenue teams like I used to run. It's the modern replacement for infrastructure that teams like mine built or bought."

"Eppo, which offers experimentation for the modern data stack. So unlike Optimizely, which focused on more kind of click through metrics, Eppo ties directly to the metrics in your data warehouse. So tying an experiment result to things like subscriptions or revenue or margins, really like board level metrics that you're trying to move."

"I think the key decision is whether you want to build or buy. ... It's usually not a zero one, it's usually both. You build some and you buy some, and it's a question of do you build 10% or do you build in 90%? I think for people starting, the third party products that are available today are pretty good."

"This product in particular, I probably spent more time with legal than any other products that I've ever kind of been responsible for. ... It is also privacy and security champions. It is, frankly, developers, like the people who are using it, listening to them. ... We're actually at a place now where we're able to partner with the Azure Department of a Responsible AI, and they've created some really extraordinary models that help detect I'll call it sentiment for lack of a better word, but basically when there is something that is patently offensive."

"AI guardrails do not work. I'm going to say that one more time. Guardrails do not work. If someone is determined enough to trick GPT-5, they're going to deal with that guardrail. No problem. When these guardrail providers say, 'We catch everything,' that's a complete lie."

"AI red teaming works too well. It's very easy to build these systems and they always work against all platforms... these automated red teaming systems are not showing anything novel. It's plainly obvious to anyone that knows what they're talking about that these models can be tricked into saying whatever very easily."

"Media Mix Modeling is now making a comeback... the company that's leading the charge of bringing the Media Mix Modeling methodology of the traditional advertising era and ushering it into the digital world is a company called the Recast."

"I finally got one that actually works. It's called the Matic Vacuum... basically, I think it's like former Google Engineers basically built like a mini Waymo car. So it has machine vision and it will avoid absolutely everything."

"From infrastructure perspective, on data tool, my first tool usually, one is some sort of data hub segment, right? This next one is some sort of a product analytics tool. Think about Amplitude. I know PostHog is actually a pretty popular one... The third piece I think that's pretty essential, I counted in the infra, is some sort of a lifecycle marketing tool."

"The success of this is you identify the gaps and eventually you want to establish something called the data dictionary... The data dictionary will include, here are all the key actions, what's the event name for each of those, and what are the property and things like that."

"I think the calculus on build versus buy is changing... because this whole space is so nascent, often your own esoteric context, your content, your workflow is really key to unlocking the power of the agent. And so I think there's value in experimenting with your own internal agent development."

"Monorepo, now for companies, it's a very much one of those door A, door B kind of things. It's a very consequential choice that is incorrect to go say yes to at a certain size, and then it becomes very correct in my mind to say yes to, but at that point it's an enormous amount of effort. So, it's a kind of thing that actually is something I'm uniquely positioned to be involved with because it's actually a business strategy thing as well."