There are so many variables here. My question is how much do you have to invest into getting it done right?
Local has come a long way, but it is still limited and slow. And while there are some people who have done stuff like this, the field is so new that you're probably going to get someone that doesn't have direct experience with everything. In other words, they're going to get stuff wrong. You will have to rebuild some part of it. You might not purchase the right hardware. Can you live with this?
In all fairness, though, if you have someone who has experience in evaluating new systems and using them to build something, then you can still be in good shape. I mentioned this, simply because it's a skill that is not as common as we would like in this world. Just look for someone with a track record of delivering functional software using new technologies.
My personal bias is that I love to keep as much local as possible, but I also realize that I bought a $3,000 machine that so far has saved me $5 in tokens from an external API. As I see it, the only real reasons to have local AI at the moment is privacy, but that does fit your use case.
As for a turnkey solution, they have their benefits, but their moat is significantly smaller now than it used to be. Quite frankly, you can vibe code the majority of TurnKey solutions in a weekend. Well, at least the parts that you need.
Sorry to not give more specific answers, but a lot of your questions may depend on whichever developer you decide to use. There's not necessarily a wrong answer in many cases, there are multiple paths to achieve what you are trying to do. If I were you, I would focus on long-term maintainability and security of your system. For example, you can have the best thing in the world, but if you can't pass a SOC2 (or, even worse, your developer has never heard of something like that) then you are going to be in a lot of pain.
These are available to read completely free online, but I do plan to purchase the physical books some day.
[edit] I realize that I should probably give more context to my answer. The books on the site are basically interviews with the authors of the software and they discuss what choices they made as well as the advantages/tradeoffs of this approach. In other words, the direct answer to your question is to learn by reading what other people have written about their own successes and glean from that.
[edit 2] Your favorite LLM could also provide a list of books that are similar in spirit, but there's just something about the series that I linked to that I like.
This resonates with me, but what helped was to come to a realization.
I love coding. I love low level coding. But I haven't written any code in the last 6 months, roughly.
But I don't feel empty or dissatisfied at all. My realization for myself is that I love solving problems and designing elegant solutions. Using AI, I still have to do that. I still have to make choices about how I want specified modules to interoperate. I know the big picture, and can catch when the AI is taking a shortcut.
I realized that it is all still low level design in my eyes. And I enjoy doing it, whether with my own hands, or with AI.
The amazing thing with AI, though, is that it has allowed me to explore so many things much quicker than I could before. It's still my design, my ideas, and even my poor choices that I have to deal with, but it reduces the boiler plate, so to speak.
That, and it has allowed me to explore more than I ever dreamed that I would be able experience.
Nice. I have a friend who is a young accountant. I have tried to get him to consider AI, but he claims that they tried it and it's not that good. I've tried to get him to understand that AI has improved dramatically in the last few months, not to mention the last few years (their point of reference, I believe).
I know a lot of accountants. One is a chief accounting officer at a medium-sized tech company and she has already replaced about 5 people in her org with AI. She says she sees a lot of low hanging fruit in finance that will be replaced by AI at her company, by her specifically. I know another partner at Big 4 that is going heavy into AI usage as well. The idea that AI isn't good in finance and accounting is a myth.
Interesting how "low hanging fruit" always stops just below the level of the person doing the fruit picking. Check back with her when her own boss replaces her with an AI, and let us know how she feels about it.
First of all accounting as a whole is incredibly broad. The fact you don’t recognise that in your post with nuance shows you have zero clue what you are talking about.
E.g llm’s are useless in tax auditing. How do I know this? My brother is a partner at pwc.
I'm in the same boat. I have 2 old servers that I let get "too" old, and now I'm afraid to touch them to update them. However, with some of the shenanigans that the Linux distributions are pulling around age verification/attestation, I'm considering bailing on them entirely.
Note, I did try Artix, but when it broke last week after a restart (in which evidently something had gone wrong with an earlier kernel update), and I had to pull out a rescue ISO, I decided I didn't want to mess with that. I switched that machine to Devuan, but the jury is still out for me. I don't have any major complaints, but I'm still in the burn-in phase. :) I'm running Arch on a laptop, but they have been a bit hostile in the community with censorship, so I'm just waiting for a free weekend to blast it and put something else on. I don't want political drama in my software.
This all comes at an interesting time, though. This is the first time that I purchased a new laptop and didn't even let it boot into Windows, but instantly installed Linux. And everything "just worked". And now that I'm excited to try Linux, so many of the big players are embracing the steps to erode privacy (AI everywhere... age attestation/verification... telemetry on by default...). It's sad, and I'm just going to "nope" out of any interactions with them.
FWIW, I once abandoned an Ubuntu server for a decade, and managed to update it painlessly in 20 minutes. That same server is still running today, now with the latest LTS. I think I might have even started with Ubuntu 4, or perhaps 6, and it has been painless all the way through. Perhaps my slow upgrades saved me from early adopter woes :).
I use Debian now much more. With all the supply chain attacks, Debian Stable feels like an absolute jewel, even if there are always a few packages I need to handle separately because I want or need a more updated version. But I love the old school no-nonsense engineering ethos.
I have a machine that's on a Debian installation that I've been steadily upgrading, one Debian stable release at a time, since I originally installed it about a decade ago now. At one point I even copied the entire installation to another disk (just a "dd" from its original SATA SSD to a new NVMe one, plus some partition and filesystem resizing), and I've upgraded the CPU/motherboard/RAM at another time, and it just keeps going reliably. It's fun knowing that the origin of that Debian installation predates every hardware component it's presently running on (with the exception of only the case and power supply).
> something had gone wrong with an earlier kernel update
That's mostly problem of Arch/Artix, they're the bleeding edge, which is not always the best for stability.
But no one said that rolling distro is supposed to always ship latest versions of everything. I've been using Void Linux past months - and while it's a rolling distro, it runs LTS kernel (mainline is also available) and maintainers are more focused on stable versions of apps than on faster updates.
My servers/VMs typically run either FreeBSD or Alpine. A Debian here or there where needed (proxmox, VPS that doesn't support Alpine, corp stuff, etc).
I've also got a couple of test systems running Chimera - going to wait until it hits stable before relying on it too much though. Experimenting a bit with AerynOS too.
I hope FreeBSD has longer supporting cycle. Its release has a supporting life of less than one year, if missing the upgrade window, then later upgrade is more difficult than others such as debian stable.
Thank you for your reply, but I haven't paid for Claude, and I don't plan to with my other subscriptions. Plus, I'm trying to have this be part of my headless automation with my own, local llms. If it weren't hacked together, it wouldn't be appropriate for this website in the first place! LOL
In anyone in the future is reading this, I ended up vibe-coding a chrome plugin to let me pull the page info as a MCP call. It's working (and I successfully can extract data) using Qwen 3.6 locally (GMKTec EVO-X2, 128GiB RAM).
It would help to have some direction of what to do on the landing page.
Editing to give more context:
I'm on an ultrawide and didn't see the help at the top right corner.
But, now I'm thinking: "OK, I typed something... what do I do with it?" Is the intent to copy/paste it somewhere? Take a screenshot? Does it convert to a less-savory format (e.g., latex)? Just looking for some directions.
Working on json, pdf, and latex downloads. Just wanted to get something out in to the world.
I've been using it for general scratchwork (hence the name), I don't really know what I expect others to do with it but I guess I should figure that out. I definitely think it could benefit from a first time prompt to open the shortcut menu.
Thanks for trying it out, I really appreciate the feedback!
I gave up trying to read the article when animations started happening when I scrolled. It's annoying and I have better things to do than waste time waiting for your animations to stutter and finish moving around when I'm trying to scan the article. Good luck with whatever you're trying to do.
There's multiple different approaches with both 4-shape and 7-shape systems being common. But the point is that your color system seems largely correlated to it, and there has been research done on the shape note system.
For the body copy, I think it's a version of Rockwell. [0] It fits the time, as well as the lower case "g" always looks quirky to me in rockwell-flavours. Stubby tail + serif on top. The heft on the headings also matches Rockwell Extra Bold with a couple tiny variations. Plus, just simply... slab serifs.
Things working against that are:
- % is wrong. That really looks like a different typeface all together. Not unheard of, might be worth seeing if it matches any other monotype fonts.
- Bolded headings have some differences. Rockwell Extra Bold should still have circular tittles, but unless it's a scanning artifact, the few lowercase "i" examples I can find in those headings seem to be square.
- The Rockwell favour in the tables is tweaked, with no descenders and uses tabular digits. This is pretty common, but the digital copies of Rockwell I have laying around don't have those exact forms... doesn't really say much when we're talking about what specific hot-metal type casts did monotype sell them 90-odd years ago.
---
On the title pages (like page 13), my best guess is Memphis. [1] The R is wrong for Rockwell, but also the lower a in "Brand" is totally wrong for Memphis, and the quote is totally different. Going to take lunch, and possibly come back to this in a bit because now I'm intrigued haha.
Comment got deleted, but Gallatin isn't the title page font. That was a digital font released in 2019, which is meant to look like Memphis with a two storey a. https://fontsinuse.com/typefaces/128627/gallatin
That does mention that Linotype had a Memphis flavour with a two-storey "a" though... so maaaaaybe it is Memphis! Most likely their Rockwell typeface was also supplied from Linotype in that case, probably under a different name.
I could be wrong, but I think that Stymie is a closer match for the body than Rockwell, particularly due to the serif on the middle arm of capital E (and F). I think that Stymie also matches other details as outlined in the comparison page on Identifont (http://www.identifont.com/differences?first=stymie&second=ro...)
Ohhhhh! REALLY good catch, I think you're right. Matches all of the features!
There's some really neat uses of it on Fonts In Use. [0]
Also a new-jersey based foundry specimen book, from a few years before publishing this catalog. [1] There's a non-zero chance that these samples were originally what convinced the original designer to go with this typeface, if we believe the designer also worked in Corning New York at the pyrex office.
Agreed, the design is really strikingly beautiful.
The typography is part of this, but I suspect you may also be undervaluing how much the overall design contributes here. The layout, use of whitespace, use of different fonts and sizes to convey hierarchy. It's just really good design made with care and attention by a skilled practitioner.
It's a great font but I do not see anything that looks similar in the PDF. This comment is more like "Speaking of fonts, I like this unrelated san serif mono spaced font."
Local has come a long way, but it is still limited and slow. And while there are some people who have done stuff like this, the field is so new that you're probably going to get someone that doesn't have direct experience with everything. In other words, they're going to get stuff wrong. You will have to rebuild some part of it. You might not purchase the right hardware. Can you live with this?
In all fairness, though, if you have someone who has experience in evaluating new systems and using them to build something, then you can still be in good shape. I mentioned this, simply because it's a skill that is not as common as we would like in this world. Just look for someone with a track record of delivering functional software using new technologies.
My personal bias is that I love to keep as much local as possible, but I also realize that I bought a $3,000 machine that so far has saved me $5 in tokens from an external API. As I see it, the only real reasons to have local AI at the moment is privacy, but that does fit your use case.
As for a turnkey solution, they have their benefits, but their moat is significantly smaller now than it used to be. Quite frankly, you can vibe code the majority of TurnKey solutions in a weekend. Well, at least the parts that you need.
Sorry to not give more specific answers, but a lot of your questions may depend on whichever developer you decide to use. There's not necessarily a wrong answer in many cases, there are multiple paths to achieve what you are trying to do. If I were you, I would focus on long-term maintainability and security of your system. For example, you can have the best thing in the world, but if you can't pass a SOC2 (or, even worse, your developer has never heard of something like that) then you are going to be in a lot of pain.
reply