Kendall Miller’s Post

An interesting chart would show the profit by market, not revenue. I suspect Mercedes and BMW make a disproportionate amount of profits in North America given the models they choose to export here and the price points. Profits matter more than revenue.

I'm humbled with the opportunity to serve on the board and grateful for the giants who have come before me. I know we've got work to do and I'm looking forward to helping do it!

When I was getting Loupe off the ground with Gibraltar Software we worked with Jason to get the most critical thing we couldn't handle done - ecommerce & payments (with FastSpring). This turned out to be a great choice because we didn't expect to get International customers, and they were ready for it (and we weren't!). Jason is endlessly creative at coming up with new ways to rev business - inside and outside of the payments industry. He's got great experience with multiple exits and PE as well. Definitely worth a conversation! (He also has some not-to-be missed sea stories)

"With Generative AI, we're able to move 10x faster than we could before; it's a huge benefit." I was talking with a friend at a major tech company responsible for migrating several processes from one tech to another. It had been slow going - and work no one wanted to do. But, once they trained a model on both sides of the problem, they could move far faster, with less effort - and keep people interested in the work. Why? Because now they just had to review and adjust the migrated configuration - which meant thinking about the interesting bits of the process, not the boiler plate. It was a clear win. This is a great example of what the current state of AI can do well: Translate between two well-defined languages to express a solution to the same problem. It works because: 🔶 There are a small number of strict rules for correct language on both sides. 🔶 The variation of source data is small (relative to, say, pictures of the world or all the written prose produced in a year). 🔶 They weren't trying to solve all the problem, just the repetitive bulk of it. If you're feeling pressure because you've yet to find the right space in your business to apply AI and you think you're alone, missing the boat - look for a space that could be described the same way as my friend's conversion process. Some similar examples I've heard working well in the past few months: 🔷 Creating RFP response drafts, particularly in the B2G space: So much of each response is the same, or nearly so, from RFP to RFP - and you want to exactly match the RFP's format to ensure a complete response. 🔷 Upgrading legacy source code: Going from COBOL to Java. The strictness and simplicity of COBOL was key to making the process work. I've heard of other attempts to go between modern languages failing due to the complexity of the source or target options. Most importantly, don't feel left out if you're not formally doing much (or anything!) with AI yet in your business: It's early days. Encourage ad-hoc experimentation, but be ready to leave it behind if it just isn't saving time or producing good results. #GenerativeAI #ProcessAutomation #FutureOfWork

Whenever you see a wicked demo of what Generative AI can do, remember that you're asking it to produce the most likely representation of the answer to your question in the medium of your choice. That's it, that's the trick. It's not that it is producing the correct answer, right answer, best answer, or even usable answer - just the most likely one given its training model. Just about by definition this means it'll look really convincing - that's exactly what it selected for. This is the biggest problem with evaluating the effectiveness of GenAI using the same toolbox we've used for years - because we're used to "looks right" highly correlating with "is right". We've taught computers cargo culture at scale.

Tech folks - any recommendations for my friend?

How sharp are the boundaries of your bubble on Linked In? I've noticed waves of common messaging in the content I see on Linked In - like herd of lemmings(1). For example, right now are echos of how if you don't adopt AI in your business your company/product is a zombie waiting to die. Frankly, I don't buy it - and I bet that sentiment is more a reflection that my network is composed of progressive engineers willing it into existence than a broad consensus of the market. Thinking naively about how the Linked In content network works, I can imagine how it would tend towards sharp edged bubbles - where you see content from people you liked/believed in enough to cross a relatively high bar (connecting) and therefore tend to think & respond like you do. If so, it's something to really watch out for professionally: ❌ If you're in a cohort that is facing an external threat - like layoffs, a down market, an external competitor - then you're likely to see it over-amplified in your feed. ❌ If you're in a market on an upswing then it's Tesla's and Pacific Island vacations for everyone - except you, who's here because you feel adrift. What are you seeing in your feed that just doesn't feel right? 1. Apparently, a group of lemmings is called a "slice", but that felt esoteric enough even my zeal for accuracy couldn't convince me to use it correctly here.

Intriguing perspective from Jason. If you knew in five years your platform would need to deliver without anyone touching your UI, how would you pivot your investments now? I'm thinking about Loupe - I could absolutely see it: People using their AI to query the telemetry we've captured about everything going on with how their applications run in the field. I think we'd have to adorn our data with a lot of the "special sauce" of analysis we embed in the UI code for it to be great.

Somewhere in the world, today, development started on a new accounting application in .NET Windows Forms. When you spend your career trying to keep up with the latest technology, it's easy to forget that the real value is just solving the customer's problems - and if you've done that with a platform introduced over 20 years ago that's absolutely fine. Loupe is here for that - the #2 enhancement request in the past six months was clear documentation for WinForms in .NET 6/8. #1 is some enhancements for .NET Containers (coming soon!)