10,126 karma · joined June 1, 2018
There's nothing quite like the Cybertruck and I don't think there ever will be again.
Knowledge workers use it primarily to "augment" their work, answer questions with AI, get it to create documents, etc. But it's usually 1 individual to 1-2 conversation threads.
Meanwhile in software development were running fleets of agents. At work we have it setup where firing off one agent request will eventually lead to like 10-20 AI calls at different parts of the process.
I make this distinctions because it changes how each group thinks about AI. For software devs it just feels like "moving up the stack". Yeah you're not writing code anymore but there is still plenty of work to do and problems to solve within your "software factory". Contrast that with knowledge workers that feel like the AI is making all of their decisions for them.
I almost think of this as a factory worker versus an artisan. The software devs are "factory workers" where we used to be on the floor but now we're managing the machines that took over the work. There are still tons of interesting problems at that stage and you still feel like you have worth. Contrast that with a knowledge worker that's more of an artisan in this analogy. For them they aren't running a factory but instead watching a robot do the work they used to do while they sit idle and watch it.
Yes it's not the same as hosting as your own hardware but I would argue that it accomplishes the same goals. The goal is to get other people out of your data. With a "service" the underlying company has every incentive to look at your data and use that to improve it's service, but when you host something yourself with a provider, it's in that providers best interest to never look at your data.
For me it all comes down to the goals of the people your are doing business with. With Cloudflare at least, their goal is to just upsell me on hosting equipment, they couldn't give a rats ass what I'm hosting on that hardware so long it's not illegal.
I'm working on a sideproject called Rowbly[1]. It acts as a sharable data store where LLM's can dump research rather than keeping it in their memory or throwing it into a spreadsheet.
At first I thought "I don't need an MCP, I'll just expose a CLI" but that carries a pretty big limitation in that it only works with agents on your computer (Codex, Claude Code, Pi, etc). For the folks on here this is not an issue and is often times preferable but I'm also targeting the average LLM user that primarily interfaces with it via "consumer AI" and with those tools the stuff you can do is very very limited.
I still think MCP's have a long way to go maturity wise and hey maybe in a few years we will figure out a better way to do things but for now, if you want to interact with consumer AI apps, there's just no way around using them.
[1]: https://rowbly.com
- The HF incident was to get the attention of investors and convey the message: "we have this incredibly powerful technology, imagine where we will be in 5 years"
- With this influencer campaign it's just a drive to get more people to use ChatGPT / sign up to ChatGPT so when IPO time comes, they can show strong user growth.
Over the past few months I've seen various reports that OpenAI's financial position isn't great and if those reports have any merit, then likely the higher ups at OpenAI are panicking that they might become a WeWork 2.0. Everything they are doing now is an attempt to justify their $1T valuation to investors during the IPO.
I don't think it's that hard of a question to answer. I've noticed on my team, our thinking has shifted from how do you directly solve a problem, to how you get an agent to effectively solve the problem and not produce slop in the process.
One thing that we have done that's probably made the biggest impact is alot more upfront architecture with the knowledge that pretty soon agents will be running wild all over the code. Having worked with these agents for a while now, you get a very good sense of how they will go about solving a problem and the various footguns they will encounter along the way. Editing an AGENTS.md file or building a skill is not nearly as fun as coding by hand but it will pay dividends over and over if you do it right.
Another big thing is doing refactoring passes. Early on in our projects our agents generated ALOT of slop and we had to go back and fix alot of it. But every time we did one of these passes, a major aspect was improving agent instructions / skills / etc so it doesn't happen again. It can be a painful process at first but I found that over time, the amount of slop the agent produces goes down by orders of magnitude.
I feel like we're still very much programming, but we're now doing it at a "higher level" where we are not writing the code ourselves but instructing the agent to. And IMHO, properly instructing an agent on a production codebase is not a trivial task.
It's like saying "Oh this car manufacturer claimed that their cars were the safest in the industry but couldn't prevent this driver from flying through the windshield" and omitting the fact the person wasn't wearing a seatbelt.
The better way IMHO is use local models so data never actually leaves the device.
Once in a blue moon it's actually the model making a material error in it's thinking and I have to go back and redo it.
Support had no answer for me other than "well you need the app". When I told them I didn't have a phone to install it on they just said "no that's not correct, you must have a phone..."
Often times it's not even that the code is bad but rather that it's overengineered. I see it happen so much that I'm tempted to actually go the other way on a toy project. Like what would Claude or Codex come up with if I told it I wanted an enterprise grade, globally scalable, compliant and auditable tic-tac-toe game.
The problem I've always had with trying out a tiling window manager like hyprland is you're going to spend a very long time trying to get everything just right. With Omarchy I get a really nice hyprland setup right out of the box.
PWA's are freaking awesome for both users and developers. Users don't have to install a massive app and me as a developer can focus on making one platform better rather than 3.
And they are pretty freaking powerful in what you can do. Not to toot my own horn too much but I created an "app"[1] for viewing cities on the other side of the earth from where you're standing. It's got camera, gyro and GPS to determine what you're looking at to the folks I've shown it too, anyone non-technical has been completely fooled and thought it was a proper app.
If I go on X, the vibe there around AI is amazing. You have folks constantly talking about cool things that they get their agents to do. Agents can help you make things you didn't have the knowhow to make, teach you things at a level that you couldn't get just by searching Google and take care of the boring parts of your life so you can spend more time doing the things you care about. Overall things are very positive and makes you excited for the future.
Contrast that with the public comments by AI executives. Pretty much all of the messages fall flat at best or are just outright doomery at worst. It's such a stark contrast and while I'm personally really excited for the future, I can totally see how someone who isn't plugged into that world to get a very grim view of the future.
And even when they run ads about AI, it's always the worst examples (thinking of that Gemini ad that talked about writing a letter to your family with AI). I really wish the industry could have a better messenger around this stuff. Don't talk about how it can help you do stuff you already know how to do, run an ad about how it can enable you to do stuff you couldn't do before.
I think the best example of this is a bike mechanic fixing a Reevo bike and writing firmware using Claude Code so it no longer required an app to work[1]. The author outright said he didn't know how to code or do anything in that area but was able to figure it out with the help of AI. I think a message like that around AI is awesome but no all we hear about is how AI will take your job.
Sol gave me a straightforward bulleted list, easy to scan and read. Opus 5 though....it gave me a solid 7 paragraphs of how it worked. I read through it and yeah it nailed the same points Sol did but the output was way harder to read.
A big part of this is ensuring that a "bad" engineers can still write solid code and also investing in systems that make it easier for us to review code.
When it comes to writing code, we have this entire library of coding standards that we've moved from one project to another. It describes, sometimes in excruciating detail, exactly how we want our code structured, antipatterns, best practices, etc.
On the review side we have invested equally into skills that split up code into readable chunks, take screenshots of any UI changes for quick validation, and a whole battery of tests to ensure that we're not generating slop.
If you were to look at just our development process, you would conclude that we're very lazy engineers. We seldom write code by hand, we seldom ask for corrections and our reviews are more of a cursory look at the PR rather than a deep review.
But the real work is not in the "development layer", it's now in the "agent layer". Making sure the agent knows how to write solid code so we don't need to write code by hand, making sure it doesn't make dumb mistakes so we don't have to correct it, and structuring our review process in such a way where an engineer only has to take a cursory look at the code.
The key difference we noticed between the "old way" and the "new way" is the "new way" is way more scalable and we're able to move way faster than we ever could before.
The reason we are talking so much about that now is that at the current level of productivity, we just don't need as many humans to do the work, as companies can use AI as leverage to not hire as many people.
This i feel will be a short lived period where not everyone has adopted AI to the fullest and thus simply having AI gives you leverage to not hire as many people and still stay competitive. The same thing happened with computers writ large, during the early days the folks that used computers had massive leverage over those that did not.
But what happens in 5-10 years once most orgs standardize on that leverage and everyone has it. All of a sudden that advantage AI gave you over others evaporates and you're right back to how things have always been, that is more humans give you more leverage.
I think once AI is broadly adopted and standardized, there will be an insane boom period for workers.
The trap here is the word "seems". When I started building my SaaS years back I thought the problem space was trivial (even joked to my business partner that I could spin the whole thing up in a weekend).
But almost 5 years on it's anything but trivial. The issue isn't with the core features, the core features in our app have been running for years without issue. What is very much non-trivial are all those subtle tiny features around the main feature that enhance the experience.
My customers don't pay me for the primary feature, that they could reproduce in a weekend. They pay me for all the tiny sub-features that came out of all those minor things that I mentioned in my first comment.
Those tiny details are the reason that it's both very easy to say things like "SaaS is trivial" and why the entrenched players are still at the top.
I'm currently building a SaaS with my business partner. I'm the "tech guy" while he has decades of industry experience with the industry that our SaaS is targeting.
What I found building this is there is a giant subset of preferences, problems, and other minor details that aren't written down anywhere, they are just kept in the minds of the people in that industry.
The reason it's not written down is to those people, these things seem obvious. And these things are both extremely important but also very subtle.
We use AI heavily in designing our application but there's always a testing pass where my partner goes over some changes and calls out subtle things of how "things work in practice". These things aren't in any manuals, they are seldom even written down and thus AI systems are completely oblivious to these details.
Now you could build a system that skips some of these subtleties and still gets the job done but what you will find is potential customers will always return to their existing system that takes these into account.
Now one day we might get an AI system that detects these subtleties but in order to do that you would need a load of training data around those things. But how do you even get that training data if all of it is held inside the minds of the people working in that industry?
Anytime I see a "1M Context Window", my brain always goes "Gotcha so a 250k usable window"