Hey I’m pessimistic about the economy, crypto and AI and such because allot of it is built on garbage and hope and it’s insane it’s not collapsed but hey maybe it never will, I guess we could just not acknowledge the insanity and just roll with it, power is because I’m working engagements actively that are doom and gloom and it’s a major theme of the work I’m approving this year.
I’m optimistic about the YoYo scene I think in person engagements are on the rise and overall even with the economy the hobby is healthy. Over saturated and due for a consolidation which will suck but that’s been happening we got a thread about the ones we no longer can buy from.
I’m personally very optimistic about my own career as I see military contracting only getting more funding in the near term and my senior level experience being more valuable as the talent pool ages and younger folks can’t find spots in the field leave me with more opportunities.
That all has baggage and it’s uhh not great but hey I can find the silver linings I made great money during the Great Recession while in the banking sector.. I didn’t feel great that I did but hey thems the breaks..
I’m just pointing out the trend is unsustainable because it’s built on hopes and prayers and gambling on a future
That is pulling plan information from sources like the company I work for. the ‘plan’ you are listing is the result of my application creating our custom set of plans and rates and applying them to a specific user and their characteristics. The parameters for each of these companies to create plans and then apply them to large pools of members is very complicated and involves multiple types of plan information and rates combined together using custom logic for each state and region. The workflow necessary to both create and manage our members in these plans is the application we have developed over the last six-plus years.
Again, you are using off the shelf components to cobble together available information sources and putting that into a form. I did not see most of the features that you prompted for and I suspect that this is simply a copy of some other freely available application for searching available healthcare options based on geographic location.
I think it is great that you are developing your own applications. Many applications of this type would never be written without AI since the relative utility and simplicity would not justify the expense of a professional developer. In that sense - it is revolutionary. Now we will have any number of new ‘apps’ to save photos, search email or organize my music library; all thanks to the power of AI in the hands of novice users. If these apps suit your needs; then more power to you. But what you are talking about and what I am talking about are completely different beasts.
Looping back around on this video now that I’ve watched it. What they are talking about isn’t really revolutionary. It’s kind of a lazy version of Spec Driven Development, which is applying specs that are normally used every day in software development and applying them to your prompts instead of willy nilly prompting with abandon.
Spec driven AI is the only way to do AI really in a corporate setting versus just making simple prompts.
Or Kiro Spec (which I have to use). Basically you start with a prompt, provide documentation and basic requirements in it and create a requirements doc. Review the requirements doc, and use it to make a design doc. Review that and use it to make a tasks doc. Finally have the AI agent follow the tasks and act on them. This gives you more control of the flow and lets you stop some dumb ai actions before it tries to do them. I used this today to create a silly amount of unit tests for a new package and update some features.
I will say this. I had apparently been on Claude 4.5 Haiku because that’s the default in Kiro and cheaper. But I realized I could manually select Opus which costs waaaaay more theoretical tokens. Being part of The Empire, I have access to infinite tokens with zero oversight though so I switched over to Opus full time and it’s much more competent when combined with specs. The new feature I set up today would have taken me 1-2 weeks and opus did it while I yo-yo’d. I mean while I did import stuff.
Can it build infinite enterprise software from scratch while integrating with a bunch of legacy stuff and passing a 2-3 month security review?… I still have so many reservations. Not to mention moral and ethical qualms. I do however enjoy health insurance and yoyo.
As the guy on the security review side of things. Lol sadly it’ll eventually pass but not cause it’s secure but because of business need and enough exceptions to policy get submitted to wave hands over the bad things and pretend they don’t exist
We shall see, I’ve been on the receiving end of months of delays while security review changes get implemented.
Plus since I got my BA in Information Security I also do security reviews off the side of my desk occasionally and I’ve handed out the same delays. Regardless of how important a project is we will stonewall folks, I’ve had the privilege of telling VPs that their important project doesn’t matter if they can’t secure it. One of the nice things about working here is security is so thorough. I’ve chatted with folks at banks that aren’t a fraction as thorough. Sometimes it gets tiring though lol. Like finish a project and are 99% done with a security review and a new surprise requirement comes out that will take weeks to implement.
I think it’s more like devs have to use AI so they will but launches will get more and more delayed as slop gets cleaned up.
My brother told my a worry too much when I asked him if he had his solar generator prepped for the storm coming. I reminded him my day job is coming up with what ifs and worrying about maybes.
Got tired of jumping between a dozen different sites and channels for tricks so I made a comprehensive trick list app. Has over 1200 tricks with links to video tutorials so far. Tricks broken down by style 0A-5A and difficulty. Some of the videos are missing, but I’m currently out of tokens for the month…..I’ll keep working on it when my token supply refreshes. I’ll probably make a dedicated post for it so folks can comment tricks and sources to add and I’ll keep building the list. Feedback would be great.
My company, like others, has locked-down the npm-registry. This has made new development almost impossible. Migration of code to new versions is also blocked. Since I am the technical lead for over ten open-source applications; this is preventing me from doing my job.
So, yes, I hate all security folks for reacting to an old, stupid registry-hack by making any new software or versions inaccessible unless I file paperwork in triplicate requesting it be stocked to our private nom-registry and artifactory. This effectively makes maintenance and development unworkable since I have to wait for days.
Still waiting to hear back anything about my gripes.
Claude Co-Work is out now. AI agents about to be used by the whole world. Every mid to large company is gonna want to run agents local. The build out and demand for physical materials will be enormous.
I’ve never done that before, that’s for sure.
