What I love the most about this beautiful reflection is how I know this will shift over the years. You can feel it already, that this is a mid point not an end state. A specific kind of clarity that will later look like only a piece of the story.
Yeah, that has certainly been true of my Mode experience, that how it felt in the moment and how it feels now is pretty different. And I'm sure this will be similar, only far more amplified in lots of ways.
My biased hope is that you take that massive corpus of learning and unique pattern recognition and apply it to an entirely different domain and set of questions. That will be powerful beyond what you can see today.
Loved this, Benn. As a huge fan of your writings, this one , as Sung Won Chung, said - was so tastefully tender. I have never worked in startups or founded one, but I learn a lot from your writings about the startup world. Now, the drive on the highway part and the poetry of it - I can very much relate. I have driven from the South (Louisiana) to Niagara Falls, NY and from Houston, TX to Santa Monica, CA and back with shorter drives such as Louisiana to Key West, FL thrown into the mix. Lots of reflections on those long, sometimes empty roads.
I LOVED this post, and am excited to see you face-to-face at the conference next week.
Startups are a long journey, and you never know how it will go. A major part of the appeal of startups is that you don't actually know how they will go. Maybe that's a big difference vs driving down the interstate - startups' road is bumpy, messy, with potholes everywhere. But it's part of the romance of working at/on one.
I hope dbt's future under Fivetran will be a good one. They both built great businesses and many in our industry rely on them.
And one day, my plan is to drive the entire I-80 from SF to NYC, and move with my family there. But that's eight years away probably.
I've done the cross country drive twice (and done a couple half-cross country trips) and couldn't recommend it more. If you ever have a chance to do it, especially if you can take some time to wander, absolutely do.
None of them? Standalone vector dbs seemed like a fairly short-lived fad. Though I could imagine there being something that solves that sort of problem, it seems like it'll be more of a mashup of something like Databricks or a model provider that gives people ways to do RAG or additional training directly.
There's a company called Yobi whose product I came across during work with one of my consulting clients. I didn't work with it directly, but it seems like something that could thrive in this dystopic climate - it's a behavioral ML model that churns out vectors of consumer data. They already have partnerships with both Microsoft and Databricks.
Yeah, I could see something like that working, that's less pure infrastructure and part application that you could sell directly to solve some business problem.
My hands-on startup adventures are now more than 20 years in the rearview mirror, but there was something poetic and poignant about this post. Thanks, Benn. I hope our paths cross again soon...
This was so tastefully tender
What I love the most about this beautiful reflection is how I know this will shift over the years. You can feel it already, that this is a mid point not an end state. A specific kind of clarity that will later look like only a piece of the story.
Yeah, that has certainly been true of my Mode experience, that how it felt in the moment and how it feels now is pretty different. And I'm sure this will be similar, only far more amplified in lots of ways.
(and also, thanks!)
My biased hope is that you take that massive corpus of learning and unique pattern recognition and apply it to an entirely different domain and set of questions. That will be powerful beyond what you can see today.
One day, this will not be a startup thing, and will be what has long destined to be, and that's a Pitbull appreciaton blog.
Loved this, Benn. As a huge fan of your writings, this one , as Sung Won Chung, said - was so tastefully tender. I have never worked in startups or founded one, but I learn a lot from your writings about the startup world. Now, the drive on the highway part and the poetry of it - I can very much relate. I have driven from the South (Louisiana) to Niagara Falls, NY and from Houston, TX to Santa Monica, CA and back with shorter drives such as Louisiana to Key West, FL thrown into the mix. Lots of reflections on those long, sometimes empty roads.
I LOVED this post, and am excited to see you face-to-face at the conference next week.
Startups are a long journey, and you never know how it will go. A major part of the appeal of startups is that you don't actually know how they will go. Maybe that's a big difference vs driving down the interstate - startups' road is bumpy, messy, with potholes everywhere. But it's part of the romance of working at/on one.
I hope dbt's future under Fivetran will be a good one. They both built great businesses and many in our industry rely on them.
And one day, my plan is to drive the entire I-80 from SF to NYC, and move with my family there. But that's eight years away probably.
I've done the cross country drive twice (and done a couple half-cross country trips) and couldn't recommend it more. If you ever have a chance to do it, especially if you can take some time to wander, absolutely do.
Which of the existing vector db data companies do you think have a chance to be the next snowflakes?
None of them? Standalone vector dbs seemed like a fairly short-lived fad. Though I could imagine there being something that solves that sort of problem, it seems like it'll be more of a mashup of something like Databricks or a model provider that gives people ways to do RAG or additional training directly.
I think you are right on the money. Unless they evolve to do more just vector DB and go deeper, they will likely be wiped out once the bubble pops
There's a company called Yobi whose product I came across during work with one of my consulting clients. I didn't work with it directly, but it seems like something that could thrive in this dystopic climate - it's a behavioral ML model that churns out vectors of consumer data. They already have partnerships with both Microsoft and Databricks.
Yeah, I could see something like that working, that's less pure infrastructure and part application that you could sell directly to solve some business problem.
My hands-on startup adventures are now more than 20 years in the rearview mirror, but there was something poetic and poignant about this post. Thanks, Benn. I hope our paths cross again soon...
The rumor is that that's the of size deal though? Or do you mean something else?