I don't think they have pivoted. They're just realising there's nothing game-specific about the database they built for BitCraft, and they're widening the scope.
The direction is more games-to-general-purpose than games-to-AI.
I don't know the answer so take this as a shitpost, but: investors and companies are throwing a lot of money in AI, not gamedev backends. AI reoriented companies are being sold with valuations in the billions.
My man, you're posting this project all over the HN comments, but I think you need to step back and really think about what you're building and what the user experience is.
Your link to an example protocol: https://dohmh.codify.nyc/protocol/71776 makes basically zero sense to me. Similarly the program page. What do you expect someone clicking on these links to take away from them?
Damn, thats stings but...totally fair so here is a basic breakdown -
1. A user would start at https://dohmh.codify.nyc and say and say whatever their intent is, e.g.: "I think I got food poisoning" or "I think I have COVID" etc...
2. The system would then either find a program for that intent or help the user make a protocol to reach their desired outcome (for that intent).
3. The program is a set of modules like: Assessments, Followups, Services etc... and the user is guided through each of them in an automated fashon. HITL is there when needed but the agent is the primary guide.
The whole point is that lots of processes just break down to a series of steps and deterministic logic that can be "codified". Thats why I call it "Codify".
I have had about 2000 programs completed but thats only 15% of the programs that users have started; so I think you may be right. I really need to improve my messaging.
One another note, I genuinely would like to thank you for the feedback. Critical feedback is hard to hear many time but its by far the most useful. I in no way meant to spam the comments but I don't otherwise know how to execute on the "tell people" part of "Do things, tell people".
Why are you using a voice recognition box instead of just a plain old search bar? It took me a second to realize what's even going on and I backed out because I ain't giving you microphone access. If people want to use voice input all phone operating systems already have that. I'd really not want to talk about my symptoms if my throat is sore from covid too.
Then I found some of the actual protocols on your website, like the "Dohmh Weight And Health." I don't understand a single thing going on on that page. Why does clicking something that looks like a video play button take me to a page with a chatbot? Why is there a list of "policy makers" on the side? I am being shown eight steps and they all look like "health Context HealthContextAgent blah blah." What am I even looking at???
I don't want to be rude but this feels a textbook example of having too many tokens on your account and too little understanding of the problems you are trying to solve.
You are NOT at all being rude. I actually feel like I am hitting the jackpot today with this and the previous feedbacks so I would also like to thank you.
I would like to provide some explanation -
0. In prison you cannot always type things on your tablet. Often, we preferred phone / voice interfaces. I am correcting this now; defaulting to text input.
1. My pops was a social service worker for 30 years; he would always explain to me all the repetitive stuff that people had to deal with and how it hurt them so this: "Codify" is meant to be automated workflows that mitigate that barrier. Codification seemed like the best way to do that; I still believe this and its not domain specific, speaking to lack of problem understanding. The agent leads people through the "codified" workflows (aka: Programs).
2. I have no tokens, budget or other; I setup my own local coding ENV so maybe change textbooks, LOL (kidding ofcorse; glut of tokens = AI slop i get it).
3. What am I even looking at??? - When you enter things like "I think I have COVID" into ChatGPT etc.. you get a wall of text that people don't want to read. When people go here https://www.nyc.gov/site/doh/covid/covid-19-main.page - same thing. What you are looking at is my attempt to turn these walls of text into real automated workflows that have agents and HITL's guide them from an intent to an outcome. I obviously am not doing that good a job ATM but 2k+m users have completed their programs w/ verified outcomes but thats less than 15% of the programs initiated so your feedback is extremely welcomed and appreciated.
Your reply to this comment got flagged dead for some reason (not sure why, it seemed like a good faith comment) so I'm responding with a couple thoughts here.
Looking at your set of steps, I'm definitely getting lost around step 3 "The program then guides you one step at a time. Each step is a small module: a short assessment (a few structured questions), a task, a referral, a verification, and so on. It adapts to your answers rather than dumping a wall of text." Am I wrong that this flow is not built yet? Or is completely inaccessible to me?
1. It comes with context in the form of paragraphs of explanation
2. It doesn't use undefined words like "program", "agent", "target". No where on that page is a string like "PersonalPrepTaskAgent".
3. It provides links to more resources, also written in nice human readable language
I don't want to be too hard on your project though. I've had to interact with the Pennsylvania DMV website a number of times, and trying to understand how to meet the requirements for a residential parking permit when I moved into my house was crazy because I needed some non-standard documentation due to having out-of-state plates. I would have loved a someone to break down the steps to me in very clear, understandable language. But I'm not sure your site is doing that right now.
As a last note, I wouldn't say you're "spamming" the comments, but I might recommend reaching out to potential users directly and asking them for harsh feedback. Posting into the void here clearly isn't getting you enough actionable feedback.
Its just not accessible to you, I am removing the login requirement right now because I think if you could easily click "Start" it might become a little clearer.
This is really great feedback man; and I am literally going to fix these things right now.
1. I am going to open up all the programs so users can click through all of the steps and really understand, like a preview.
2. I am going to just use my horrible writing skills and do the copy myself. I am very insecure about my writing so I leaned on AI for that probably way too much.
3. I am going to add a video NOT of program/protocols but of the actual process that its automating; e.g.: https://wiki.codify.nyc/agent/posts/nyc-311-apply-victim-com... and some of the user testimonials; for example - the core use case is care plan automation. 1k of the 2k users I was speaking about all were from crohns.ai
I think I not only missed the mark here but did the exact opposite of what I wanted to. Codification is supposed to simplify complex workflows and it seems like I did the opposite here and I can totally see how nobody would ever be able to tell that I am automating a process here.
The amount of embarrassment you just saved me I cannot thank you enough for. I was about to do my first "email/blast" style marketing campaign but I am going to dig into this first. If you don't mind, I would like to email you when I made the fixes.
Edit: The thing I don't know how to express is that the page you like, that can take a lot of work. I want to automated all of the things possible so that instead of reading a page its essentially just a program and the agent guides you (for real) through everything required to complete whatever process or flow the page describes.
Glad the feedback was helpful! Feel free to email me at either email in my profile. I think your next steps laid out here are good ideas. In particular, fully documenting (with video or text) a handful of actual successful uses of your site / the programs seems like a great next step.
It's because the author is comparing that incident to how they were prepared with the bulk of their analysis in advance. Ready to publish as soon as it happened.
The downvotes are because the discussion isn't really about "which option do users prefer". It's actually about returning user choice. As the original comment said: "If I want a separate tab, I can Cmd+click and Browsers don't have the reverse option for opening in the same window."
So when you say "Nope!", you're being downvoted because you're implicitly saying "actually users don't deserve choice".
The ball appears to go high enough but it [control] doesn't like that it went over space. So the best one could do was put it (after +1s) into the port.
This is one of a bunch of issues that have been popping up since a vibe coder took over the bulk of development on this project. There's a (probably also AI generated) list of a big portion of the issues here: https://github.com/benbjohnson/litestream/issues/1221. That proposal has been open for a few months, and it seems (from my POV) unlikely to be resolved any time soon.
I've been following litestream for a while, and it seems like the project has been hijacked by a vibe coder. I wouldn't trust it for critical tasks anymore.
Yes. To me, it is. Sometimes queries I give it are 100-200 lines long. Sure, I can solve it eventually but getting an "instant" answer that is usually correct? Absolutely priceless.
It's pretty common for me to spend a day being stuck on a gnarly problem in the past. Most developers have. Now I'd say that's extremely rare. Either an LLM will solve it outright quickly or I get enough clues from an LLM to solve it efficiently.
You might be robbing yourself of the opportunity to learn SQL for real by short-cutting to a solution that might not even be correct one.
I've tried using LLMs for SQL and it fails at exactly that: complexity. Sure it'll get the basic queries right, but throw in anything that's not standard every day SQL into it and it'll give you solutions that are not great really confidently.
If you don't know SQL enough to figure out these issues in the first place, you don't know if the solutions the LLM provides are actually good or not. That's a real bad place to be in.
Have you ever read Zen and the Art of Motorcycle Maintenance? One of the first examples in that book is how when you are disassembling a motorcycle any one bolt is trivial until one is stuck. Then it becomes your entire world for a while as you try to solve this problem and the solution can range from trivial to amazingly complex.
You are using the term “hard problem” to mean something like solving P = NP. But in reality as soon as you step outside of your area of expertise most problems will be hard for you. I will give you some examples of things you might find to be hard problems (without knowing your background):
- what is the correct way to frame a door into a structural exterior wall of a house with 10 foot ceilings that minimized heat transfer and is code compliant.
- what is the correct torque spec and sequence for a Briggs and Stratton single cylinder 500 cc motor.
- how to correctly identify a vintage Stanley hand plane (there were nearly two dozen generations of them, some with a dozen different types), and how to compare them and assess their value.
- how to repair a cracked piece of structural plastic. This one was really interesting for me because I came up with about 5 approaches and tried two of them before asking an LLM and it quickly explained to me why none of the solutions I came up with would work with that specific type of plastic (HDPE is not something you can glue with most types of resins or epoxies and it turns out plastic welding is the main and best solution). What it came up with was more cost efficient, easier, and quicker than anything I thought up.
- explaining why mixing felt, rust, and CA glue caused an exothermal reaction.
- find obscure local programs designed to financially help first time home buyers and analyze their eligibility criteria.
In all cases I was able to verify the solutions. In all cases I was not an expert on the subject and in all cases for me these problems presented serious difficulty so you might colloquially refer to them as hard problems.
In this case, the original author stated that AI only good for rewriting emails. I showed a much harder problem that AI is able to help me with. So clearly, my problem can be reasonably described as “hard” relative to rewriting emails.
What happens when these "AI" companies start charging you what it really costs to run the "AI"? You'd very likely balk at it and have to learn SQL yourself. Enjoy it while it lasts, I guess?
I work with some very complex queries (that I didn't write), and yeah, AI is an absolute lifesaver, especially in troubleshooting situations. What used to take me hours now takes me minutes.
Personally, as a DM of casual games with friends, 90% of the fun for me is the act of communal storytelling. That fun is that both me and my players come to the table with their own ideas for their character and the world, and we all flesh out the story at the table.
If I found out a player had come to the table with an LLM generated character, I would feel a pretty big betrayal of trust. It doesn't matter to me how "good" or "polished" their ideas are, what matters is that they are their own.
Similarly, I would be betraying my players by using an LLM to generate content for our shared game. I'm not just an officiant of rules, I'm participating in shared storytelling.
I'm sure there are people who play DnD for reasons other than storytelling, and I'm totally fine with that. But for storytelling in particular, I think LLM content is a terrible idea.
