My payment agent is named George, not stripe-agent | Hacker News Reader

My payment agent is named George, not stripe-agent

blog.kestrelsnest.social

51 pointsfortyseven44 comments

Please disclose AI use, or the name of your "writing agent" at least, so I can know to skip the article. So much "it's not X it's Y" in this post, I'm losing it.

> This isn’t whimsy; it’s how I remember who the work is actually for.

> These aren’t chatbots with personalities; they’re specialized configurations I invoke by name to focus my intent.

> That’s when I realized the naming wasn’t a quirk. It was a practice.

It is a quirk

> I’m not asking for a generic security scan. I’m saying that I need to look for what I missed.

You aren't asking for a generic security scan? It seems like you're asking for a generic security scan.

> I need to look for what I missed. I need to find the secret traveling farther than it should, the data leaking where it shouldn’t, the assumption I made that an attacker won’t make. I need to be paranoid on behalf of the users whose data and trust I’m protecting.

> The names aren’t just labels. They’re invocations. They shape my intent before the work even starts.

They are just labels.

At least right now it's mostly in AI-related articles. Scroll any AI article and have a look at the number of topic headings as well as how many start with the word "The". I have my defenses up on any AI articles and can quickly avoid the are LLM-output with aesthetic clues. An upfront disclosure would of course be better.

Unfortunately other topics are still catching me off guard, like the article about complex numbers posted today which I managed to get through a third of before realizing all the grating bits I was reading were because it was from an LLM.

literally anthropomorphizing AI agents.

To be fair, I certainly name my tools. But I didn't have to use AI to invent a whole bunch of "personalities" for them.

I appreciate the writer actually taking the time to explain why `george`. I have worked in some projects where some thing-a-majing or another is called `valhalla` or `thor` or something or another but there is no documentation as to why it is called that and the people who were responsible for naming them so have already ridden into the sunset. If I ever meet him, I "just want to talk" to this CTO who named US East region 2 as "eu2".

> I "just want to talk" to this CTO who named US East region 2 as "eu2".

How? Logically I don't get it.

Not OP... all I could come up with is they didn't remember US east vs east US, so landed on EU2 meaning 'east US 2.'

Estados Unidos 2

Consider multinational orgs - "EU2", and collisions with English when speaking "you too".

How does one person talk to another person who named a thing? Well, you can either meet in person and use your mouths, or you can pick up a phone....

(I'm genuinely confused by the "How?" question)

Eastern US 2

It's the sequel to EU. EU 2.

But EU5 has been out for over a month now

When will EU6 release be? :D

The problem is that, in any organization past a few people, someone will eventually wonder if they were the inspiration for a particular name, and not in a good way, or someone might introduce politics or something else divisive.

It's better to have arbitrary names that are memorable in some way but not common enough to be associated with someone living within recent memory.

IMHO, YMMV, yada yada

> someone might introduce politics or something else divisive.

Reminds me of a project I was peripherally involved with many moons ago. The codename for the project was "Tardis" from Doctor Who. No problem there. But we ended up having to redo a significant portion of it later, and someone had the bright idea of changing the redo codename to "ReTardis". It was hilariously juvenile at the time, but I could see how, decades later as society gotten less tolerant of that kind of humor, the codename probably has become objectionable.

Maybe I'm bad but I find this really funny in 2025

The only reason I can think of is to not duplicate AWS's "us-east-2" region name

> The tech industry loves to abstract away the human. Users become “MAUs.” Problems become “pain points.” Customers become “conversions.”

The LLM loves to torture concepts into statements with pithy veneers and three-item. Punctuated. Lists. “Pain points” as an example, really? All of these terms are just more specific than the ones they’re contrasted with, which don’t have much of a human element to them to begin with.

The irony of bemoaning this while AI-mimicking a team of people and getting a computer to write for you in its own voice…

George, Ray, Agatha... Ok As long as you are the only one managing these systems. But the moment you involve other people, this is the worst naming possible.

If a new hire can handle "Talk to George, he's our security guy" then "Talk to George, it's our security LLM" shouldn't be that much harder?

I think the natural expectation is that someone named George is a human being.

So? It's a little eccentric, but plenty of people give names like this to their computers, cars, boats, pets, etc. and no one seems to struggle with that.

Or a monkey. Or a car (when I was a kid one of our cars was called George.)

Human names need meta data. People see the job title of "head of security research" in Workday or Slack.

For new hires (or people in other orgs), shouldn't need long product descriptions trying to explain team lingo means.

Maybe at a large company that's the case, but plenty of mid-sized offices will tell you "oh, everyone knows Susan is the one you talk to about security code" even though Susan's title is just "programmer" like everyone else.

It's not like a list of six LLM sub-agents is difficult to hand out, and there's even a public blog post detailing the names, specializations, and rationale for this in case you somehow forget and can't just /list-agents or whatever.

all big companies used to be small companies. I once worked at a company where the original developers thought it would be cute to name everything star wars themed.

Give George a phone number then and let me call him on the phone.

If they have non-descriptive human names, they should behave like people.

- Our payment system is down - Call George on the Phone and ask him to fix it..

I dunno, I think it works in any organisation small enough to only have a small number of any given thing. One you start having fleets of servers then you’ve got to switch to fleet naming.

There are only two unsolved problems in computer science. Cache invalidation and naming things. And off-by-one errors.

As the end of the article says, to the author this is more of a "ritual".

I don't know how effective it is, but I can't imagine this would undermine the quality of the output, so if it adds a little bit of humor and human-ness to my workflow, I'm happy to try it out.

> I added Helen to my roster just this week.

yeah. cringe. I debate with myself is this Aghata can truly be trusted to be > I need to find the secret traveling farther than it should, the data leaking where it shouldn’t, the assumption I made that an attacker won’t make. I need to be paranoid on behalf of the users whose data and trust I’m protecting.

at the end of the day its still an llm. but hey, I want to call Claude _Claudius_ all the time but I don't cause it'll shut me down real quick

When Copilot was first released I called it Bing and it got mad. :P

I haven't written myself a Claude agent nor a skill nor a plugin yet, but when I do, I'm going to name it well.

As I've been asking Claude to "keep planning criticize ultrathink" very often and repeatedly, maybe I'll make a planning agent, one that helps me shepherd each plan well.

I'd be very curious to see what sort of code / prompting goes in to these agents, and what sort of results you see from them - is the name just a personal reminder, or do the LLM subagents incorporate these philosophies? What sort of behavioral changes do you see from this method?

It's a good way to name your agents, who do they help/work for... Smart move

Dehumanize humans, and humanize non-humans.

Makes sense to me…

But seriously, naming things is always a sticky wicket.

I tend to name my various devices as characters from Glen Cook’s The Black Company.

My iPhone is Thai Dei, my iPad is Soulcatcher, my Watch is Goblin, and my Mac is Mogaba. It helps me to keep them distinct from my simulators.

If I wanted really crazy names, I’d use Garret P.I. As a source.

First in-the-wild reference I've seen to some of my favourite books. I feel your watch as Goblin makes more sense if it's stuck around with you for a long time and generally works but is a bit of a pain to use. Thanks for the share.

Great books! Strongly recommend for anyone into fantasy stuff.

And they are still coming!

I just finished Lies Weeping, which is #12, I think. There’s 2 more on the way. I suspect they are already written.

> This isn’t whimsy

Uh, yes it is? It's just whimsy with an explanation. Long live descriptive, preferably short, names.

It's also a reminder - we're not just here for the surface concept of X, we're here for the deeper philosophical reasons of Y and Z. The goal isn't to check off a "disability accessible" checkbox, it isn't even to "think how disabled people might use this" - it's to be actually accessible to all the actual people with actual disabilities.

Trust me, there are a a LOT of people who need this reminder.

I'd expect the difference in prompts produces significantly different LLM outputs, too - tell an LLM to check boxes and it won't show much initiative, but give it a philosophy and it will often suggest ideas you missed.

Yes, this exactly! Thank you for picking up what I was laying down. I gave them these names as a reminder to myself, the person who is using these tools, who I'm doing this particular task for, why I should remember to look at what I do through particular lenses, and how to get the best output from my tools of choice.

They're my tools in my toolbox for my code. For most of the projects I've used them in, I've been the sole developer and the issues other folks have raised about naming schemes don't apply. I've shared them with colleagues, but if they use them they can call them whatever they'd like -- I'm not trying to say my way of doing things is better for everyone, but it is better for me. And maybe could be for someone else too.

(Sorry for both a three item list and a "not X, is Y" phrasing in my reply. Oh jeez -- and an em dash too. I'm working on moving my writing style away from what LLMs are throwing out there right now, but it's slow going.)

George? IDK

Reply on news.ycombinator.com