Play Your Hand
I was watching a video today where a guy explained how to better tune up the algorithms required to run a vector database of any real consequence for an LLM to use. If most of that sentence didn't make a lot of sense, join the club. I got what the guy said, but I also had no ability whatsoever to do anything with that information directly. That's what got me thinking about one area where we tend to muck things up.
We Are Who We Are
Where I work, we've got well over 100 apps that make other software work better. We talk about it like this: "do the work you want to do, only better." You know when you're on a site somewhere trying to do something, and you shake your fist a bit and say, "These headings are fine, but I wish I could number them" - you know that feeling? We sell Numbered Headings.
That's who we are. We make things better. We "make work flow." <- That's our motto thingy.
So I'm telling you all this because it sets up the idea I'm thinking about, and I got there by way of that video I watched about vector databases.
Let Me Explain This Real Quick (You can skip this subsection if you want)
Because "data" has changed to be a very mixed bag - unstructured text, videos, pictures, audio, etc - we needed a new/better way to absorb all this information and know what we had.
In the old days, we'd stick a picture in a database, and then a human (a frail weak paltry human - shh! Quiet, Robot!) would type in things like "cat, orange, garfield, photograph, feline" and on and on.
Vector databases do something crazier. They have an image analyzing technology that turns the information in the image into arrays of numbers that mean something to a computer, instead of the actual picture.
Things like ChatGPT and other natural language processing technologies don't "know" things, they "guess" things based on what information they've been able to glean. A vector database is a much smarter and more useful (and faster) way to get lots and lots of complex weird information into something reusable by a large language model (LLM).
Back to Playing Our Hand
So, the type of people required to make smarter vector databases and smarter information to feed AI tools are different than the type of people who come up with much better automation scripting tools. The people who think about deploying smarter vector database technologies are not the people who think about tracking your time across various ecosystems so you don't have to do it the old fashioned by-hand way. (Is there anything more annoying than logging time manually?)
Thus far, I've been talking about Appfire where I work. It could be about you, though. Or me.
For instance, I will never be a coder. I lack pretty much all the basic pre-requisites. In fact, I can't even direct people who make code. I can help the people who direct them. I can assemble the Avengers.
领英推荐
That's my hand. That's what I can do. It's my part.
Others can code, but maybe aren't confident with leadership. And they might be incredibly creative, or far more on the engineering and thorough side. Some people know how to close every loop, while others can type out into the "what happens if we..." sphere. Some are both, but not always.
Save Time - Work From Where You Are
People chew up time trying to be what they're not. It's like being a dolphin upset that you can't fly.
It's important that I add something: I said "work from where you are," and not "from where someone else thinks you are." The number of people sitting in the wrong bucket because of the someone elses in the world is depressing.
Are you a very organized person? Great. Are you the kind of company that only makes new things, but needs to hand those off to the company that can grow and expand those things? Perfect. Are you a writer? Or are you more of an editor?
Be the you that you are. Play the hand that you have. If your company is well suited to one kind of work, don't worry so much about chasing other kinds (unless what you do is a fading industry).
It's That Flow State Thing
You hear about flow. Maybe you've read the book. The universe tells you which thing you are, which hand you have.
When I try to watch videos about vector databases, I get sleepy if they go deep. If they can skim the surface, at least at the start, I can follow along. This tells me that I'm NOT the kind of person who should work on this.
If I watch a video about someone turning something creative into something bigger, like visualizing a wild idea and turning it into a business, then I get excited, write lots of notes, seek out even more information, and gather up a lot. I squeak out all the information and share it with you and force you to get excited about it, too. That's how I know I'm in flow with a certain kind of work and how I'm not meant for other things.
When you know what hand you've been dealt, you can understand how to play what you have.
I'm in Las Vegas visiting my parents before a big industry event next week. Meeting with smart people and praising their wonderful accomplishments is one of my favorite hands to play. And I try to play it well.
What about you? What kind of hand do you have? What about your business?
Tell me more.
Chris...
Building for SMBs *BrB*, AI+Human Content, and Automation | X - Credit Karma, LendingTree
11 个月This was actually a very good newsletter. Thank you for sharing (and not selling).
I help companies make MORE money with less! | Copywriting & Marketing Strategy
11 个月Interesting perspective
Dragon of the West
11 个月During the alpha-go iteration of AI I was tokenizing corpuses and what I loved about vectorized words is how they relate. The vector for "man" is close to the vector for "king" and the vector for "woman" is close to the vector for "queen," similarly the vector for "king" and "queen" are close but different from woman/queen (the semantic encoding of gender and role/status). Really interesting stuff. Imagine having your role be to think of that idea and make it real?