Hacker News Reader

> the models are improving at a rapid rate. If asked ~3 years ago where the state of the models are today, it would sound like sci-fi

Absolutely true, many things will continue to improve in significant ways. However, if we look at the modern history of rapid disruptions driven by technology (a side interest of mine), persistent patterns emerge. Similar to avalanches or flash floods, such periods of very rapid disruption are often triggered by one or more significant breakthroughs in certain technologies. Early rates of change tend to be fast and furious but eventually begin to taper as recently unlocked low-hanging fruit is harvested and those racing through newly found terrain encounter all-new significant barriers and points of friction. Early in such periods, extrapolating the recent extraordinary rates of change forward has poor predictive power. Sudden extreme bursts tend to regress back toward the long-term trend line.

Arguably, the current disruption in LLMs can be traced to post ~2010 research slowly building to the 2017 transformer paper and the adjacent work it quickly inspired. So today is, arguably, mid or late-ish in the LLM rapid burst phase. The rate of fundamental, broad-based breakthroughs lifting all LLM applications has clearly slowed with many of the most impactful recent discoveries being in scaling, optimization, tuning and productization toward specific domains. That doesn't mean there can't be another transformer breakthrough tomorrow but, historically, black swans rarely travel in flocks.

Progress happens in a series of S-curves. While your observation is correct that advances occur initially rapidly then taper off, the next step tends to arrive sooner than the previous, and with greater magnitude [1]. Tim Urban's article from 2015 has a great explanation of this phenomenon [2].

[1]: https://ourworldindata.org/technology-long-run

[2]: https://waitbutwhy.com/2015/01/artificial-intelligence-revol...

This is of course true in general. But the question is not "how with this evolve" but how will we deal with the rapid changes in the industry? I suspect a long term k-shape salary curve, even worse than today, with the lower 80-90pctile salaries bottoming out such that many have to exit the industry to make ends meet. You can laugh and blame them for not saving as much as they should, but that's still a fairly horrifying prospect for most of us.

I think a _lot_ about stock trading a profession vs algorithmic trading. It was brutal - suicides, many pivoting out to doing car dealership-style work. Probably a 1/10 or 1/20 survivor rate every couple years, with almost all of it a very painful five year period.

And it was the dumbest and least valuable stock traders that exited the industry. The industry is alive and well today.

Phew, for a second there I felt bad for them!

That really depends on how you define alive and well. There are still stocks and there are still traders, but the market valuations are obscene and it sure appears that there must be collusion or corruption driving the industry to jam massive IPOs into every index and 401k they can find as fast as possible to fasciliste and exit.

I would ask for references for the suicide claims, so others can assess the impact themselves. That's a very serious claim to provide without any proof, especially to a group of people who very well be going through the same thing. I am not saying it did not happen, only it's the right thing to do.

> The rate of fundamental, broad-based breakthroughs lifting all LLM applications has clearly slowed with many of the most impactful recent discoveries being in scaling, optimization, tuning and productization toward specific domains.

To me it definitely feels like it's still accelerating, with the most impactful recent discovery being RL training reasoning models (late '24, early '25).

There's an interesting article called "sigmoids won't save you" https://www.astralcodexten.com/p/the-sigmoids-wont-save-you which argues that (unless you have privileged information) you should always assume a process will continue about as long as it’s continued already. (Lindy's Law)

With that in mind the current disruption should last another 10-15 years (assuming it started in '10 or '17.)

> The rate of fundamental, broad-based breakthroughs lifting all LLM applications has clearly slowed with many of the most impactful recent discoveries being in scaling, optimization, tuning and productization toward specific domains.

What this means is that the disruption across industries not even truly begun, because it's not the generic chatbot models that are going to kill labor, it's all the domain-specific applications that leverage those models to perform work that was performed by humans

Why would it stop with just developer layoffs? When software companies rely on LLM providers to run their business, I’d argue we‘ll see a massive bust of these companies around the world - from on-prem products to SaaS.

Customers may build the software they need entirely in-house or via prompt-engineer consultants, without the need to buy software tools like today. It could be a very very different world.

This won't happen in most cases because the valuable thing is largely the knowledge encoded in the software, which the buyers of the software don't have and don't want to have since they're focused on their own business.

There's also, of course, the not insignificant value in the software itself actually working, being operated, being updated when necessary, all of that. Again just extra hassle no business will want to shoulder when they can just buy something that does it for them.

Why would I build my own CRM instead of paying 50k a year or whatever? The engineer plus tokens for maintenance will cost you way more than 50k.

These people are delusional and just repeating delusional vibe coder tweets.

Not every business has $50k for a CRM

Hell, there is a lot of really good open source software that fits most peoples needs already, that can be self hosted and costs nothing but the running of it. But people still pay for the SaaS product. Because you're not just paying for software. you're paying for support, uptime, compliance etc. These people think that SaaS is dead confuse me.

agreed. Also, data security, data compliance, legal, customer support, operations... Yeah, SaaS is not going anywhere soon.

> Customers may build the software they need entirely in-house or via prompt-engineer consultants, without the need to buy software tools like today. It could be a very very different world.

Already happening. I know of a few places that have gotten such large gains from LLMs that they know have their engineers working on creating homegrown ports of popular services (Docker etc.).

It seems to me that Docker should be far, far down the list on services to recreate in-house with AI.

But since it's open source the AI is trained on it so it can actually do it :D

Why would you create a homegrown port of Docker? Docker the container software, or Dockerhub the image repository? This is just confusing. If you didn't want to use Docker there is a perfectly good well tested alternative called Podman with wide adoption.

Homegrown ports of calendly or jira seem feasible, and arguably a good business decision. Homegrown versions of docker seem ridiculous as a starting point, even if its possible to do today there is much lower hanging fruit to go after first.

Oh. Oh my. In-house Docker gives me hope for my future as a cleanup consultant.

> I know of a few places that have gotten such large gains from LLMs that they know have their engineers working on creating homegrown ports of popular services (Docker etc.).

Sounds like a good way to eventually erase those gains.

Not sure about Docker (lol) but stakeholders are definitely more open to "building your own" now. It used to be that to be agile as a business you would seek out already built software and rent it, as it typically was cheaper than building and maintaining your own (I say typically due to stuff like vendor lock-in and such). But these days, and especially in 2026 with the widespread use of agents and harnesses, that formula has started to change. Even though the SOTA models are really good now, it's the harness and the "fluff" around the model that makes it a game changer. The developer is no longer the one writing or even gluing the code together, the harness does that. Pair that with context preserving mechanisms and tools that emerged (automatic context compaction, AGENTS, TOOLS, MCP...) and you can get to a state where you start a new thread in Codex and it knows your systems, your dbs, can smartly explore code it doesn't know and db data patterns etc., it can explain stuff to a new developer (and be correct most of the time and have time to spend on the developer)... all of which SIGNIFICANTLY reduces the risk you take on yourself as a company when you "build your own". What's $10k/year to any half-working semi-profitable company? Nothing. But in 2026, you can build and maintain A TON of software for that, much more than your "average IT needs" company may ever use.

I'm sure the very large (and very small) businesses will keep their absolute need for (or the lack of) inhouse developers, but everything in between will probably get compressed to one or two inhouse architects in direct contact with the stakeholders and the rest will be contractors working with Codex-like automation.

Nothing corroborated this. Performance on benchmarks has practically leveled off. The big gains have come from architecture (have a secondary LLM review output) or searching the internet. Also prices are going up. Everything points to the likelihood that we're at the top of the curve.

Most benchmarks can be trained for as well, so they are over-representative of model's engineering skills. The entire nature of a benchmark is collapsing some qualitative work (software engineering task, architecture choice, code quality) into a quantitative score which can be optimized for.

> Performance on benchmarks has practically leveled off

Ehm, no? DeepSWE[1] for example shows that new models like gpt-5.5 continue to show big improvements compared to older models.

> Also prices are going up.

Prices for frontier intelligence have gone up, but prices for the same level of intelligence have gone way down (what you can get for pennies now was SOTA just a couple of years ago). The pareto frontier is still expanding.

[1] https://deepswe.datacurve.ai/

> Nothing corroborated this. Performance on benchmarks has practically leveled off.

[There is plenty of data to support the claim that AI continues to improve, even exponentially.](https://epoch.ai/trends)

As for benchmarks I feel compelled to remind you that as soon as a metric becomes a goal, it ceases to be a useful metric. The models optimise for solving the benchmark and we create new benchmarks to assess broader intelligence. As models converge on 100%, progress obviously slows. That doesn't mean intelligence isn't improving fast. It just means that that benchmark is being well served and we need other benchmarks to assess other forms of intelligence.

I would like to take your bet that we're near the top of the curve. I take the side of Geoffrey Hinton, the Nobel Prize laureate scientist known for his work on artificial neural networks. He believes AI is getting better even faster than he predicted. He estimates that every seven months AI becomes able to handle tasks twice as long.

> [There is plenty of data to support the claim that AI continues to improve, even exponentially.](https://epoch.ai/trends)

This doesn't look at all exponential to me: https://epoch.ai/benchmarks?view=graph&tab=eci. OpenAI models went from 137 ECI to 159 ECI over about a year and a half, and the trends are similar for Anthropic and Google. These things have never been exponential.

> The models optimise for solving the benchmark and we create new benchmarks to assess broader intelligence. As models converge on 100%, progress obviously slows.

We are nowhere near 100% on important benchmarks like hallucinations: https://artificialanalysis.ai/evaluations/omniscience?model-...

...also, progress isn't improving with model releases.

---

We're running out of money. While we don't know how much it cost to train things like Claude, most (all?) industry reports indicate that a significant gain in function (2x) would require an exponential amount of resources (20x). No one's yet been able to convince investors that's worth it.

Also, we're running out of data: https://epoch.ai/publications/will-we-run-out-of-data-limits....

Also, we're running out of of low hanging fruit: "We find that the level of compute needed to achieve a given level of performance has halved roughly every 8 months, with a 95% confidence interval of 5 to 14 months. This represents extremely rapid progress, outpacing algorithmic progress in many other fields of computing and the 2-year doubling time of Moore’s Law that characterizes improvements in computing hardware (see Figure 2)." (https://epoch.ai/publications/algorithmic-progress-in-langua...). Maybe you think we'll continue along this breakneck pace, but again no investor thinks that, which is why prices are going up (investment is drying up).

Also we're running out of compute. Data center projects are stalling. Some of this is spiking energy prices, some of this is politics, much of this is grid constraints and supply chain problems: https://tech-insider.org/us-ai-data-center-delays-cancellati....

---

Finally, and perhaps worst of all, despite unprecedented investment data on the productivity gains is mixed. This is the biggest difference from other technological leaps like electricity, the industrial revolution, literally fire, etc. Those things were immediately, undeniably more productive. AI is not like that. You're not seeing an AI Microsoft, an AI Salesforce, an AI Oracle, an AI SAP, etc. You can argue that their advantages are structural, but there are no successful AI-powered alternative products (no AI Office, no AI ERP, no AI database, etc).

That's what the AI comapnies would like, but they can't pay back the 100's of billions they are blowing without 10,000x the price they charge. The investors won't allow it. we're not even in a revenue cycle yet and they are already trying to dump their deep losses on retail by trying to IPO

I would highly encourage you to watch this short clip

https://youtu.be/5eqRuVp65eY?si=3fLT6S5q2OIUcu6r

That is a great video, but short is very subjective, the video is 24 minutes long :))

What is your theory of when AI gets to 100%. PMs and business analysts build all the software? Or just like a 700 or so 1-founder companies in the world and everyone else is without work? The matrix?

Why would you need a pm or a biz analyst

Better question, why would you need a CEO?

RACI

> Or just like a 700 or so 1-founder companies in the world and everyone else is without work?

This. But instead of 700 it's more likely that everyone will be a founder (more or less). It's already scary how easy it is to launch an MVP or produce prototypes with the latest models.

Until an LLM becomes better at coming up with products people want and running a company. Then we'll cut out the founders and have LLM run companies.

> It's already scary how easy it is to launch an MVP or produce prototypes with the latest models.

No it isn’t. The things that were hard are now harder. The things that were comparatively easy are now easier. But if you build another piece of vibe-coded crap in a world awash in vibe-coded crap, you will not stand out. Nobody cares about your unpolished, one-shot prototype, so cranking them out faster is not really helpful.

Differentiation is always a problem of effort and care, and this isn’t going to change.

Then why don’t we see a bunch of amazing software for sale that solves every niche issue?

In my own (admittedly limited) experience, 2 employees in my company (that had no programming knowledge or experience) have vibe coded apps that simplify their daily roles. The apps basically automate a flowchart of steps where multiple people need to submit certain pieces of info and as they do, a "project" moves through stages and the employees get notified on Telegram.

The app really is just several simple forms with some if/else logic, but claude code allowed them to get the app up and running and deployed on vercel's free tier, and it's Good Enough™ to save them an hour or so each day lost in messaging and chasing up things.

I don't think anyone would ever have targeted an app for sale to them, and it would have been hard to twist some sort of flow management app and integrate it with Zapier or something to handle external api calls. With claude code they could just tell it what they wanted and solve their very niche issue. That's why I think that even though LLM coding has improved so much you might not see more software for sale - it's easier for people to just...make their own software.

Don’t get me wrong, :) that’s pretty cool! I’ve also made highly personalized mini apps for my own personal life. Currently working on an iOS one to log mood and correlate it with HealthKit data since the native health app does a bad job of it.

That said, I meant more like production grade apps that have to serve N>1, which is IME where the hard part LLMs suck at comes in. I saw a tweet somewhere along the lines of “CEOs/execs are so divorced from the last mile effort that they are uniquely susceptible to believing AI can replace engineers end to end”

The best part of this workflow - which I see often - is that by having someone build custom software to automate some process they often step back away from the process being their job. That eventually translates into them understanding that some (or sometimes most or all) of that process is not needed. There are so many corporate processes that were implemented and then become the way... and if there are people who identify that process as being their job those people resist attempts to optimize that process.

I have seem several people use AI to write apps to automate a process and along they way finally ask the question 'do we even need this process?'.

Regrettably this does not happen everywhere.

To whom will all those founders sell their AI generated products?

See my issue with these comparisons is that it always compare AI to 100%.

When 100% does not exist. Most software out there has issues, bugs, compliance problems, security weaknesses, scaling, redundancy, availability issues...etc. A lot of this is not actually related to how good or bad software engineers are. It's about costs and time to ROI. Greed is an issue too.

So people seem to have this idea that software created by humans is perfect (its not). And that deterministic (human created software with if/then) is alway going to be better than probabilistic (LLMs). Which in a perfect world is the case, but we live in a capitalistic world where this is not the case.

I don’t think that’s the comparison.

The point is when you pay for human made software from a saas, those problems are not your problem!

You get the saas company to fix it, and it they won’t you go to one of their competitors. If you reroll in house you’re now responsible for every bug

>If asked ~3 years ago where the state of the models are today, it would sound like sci-fi if answered, "the models are creating full MVP apps in ~30 minutes with one prompt".

The first one-shot app was created with ChatGPT in June 2023 - 3 years ago. In my experience, the current result of one-shotting apps is just as bad today as it was back then.

What “full MVP app” are you talking about? I know of none that have been anywhere near production ready. With all due respect, I think you’re portraying fantasy as reality. I would love to be proven wrong.

> The first one-shot app was created with ChatGPT in June 2023 - 3 years ago. In my experience, the current result of one-shotting apps is just as bad today as it was back then.

Hard disagree. Take a good SaaS starter template and do a bunch of harness engineering. You can get an agent to shit out production grade stuff. You might argue that's cheating, but there's nothing stopping you from doing it, and it works. It's only getting better too.

The original comment asked for someone to name one and you didn’t, though.

Cos I'm too busy doing it at work.

Can you share one when you get home? I'm interested to see some examples.

Send your agent off on a task and give us an example come on now

> In my experience, the current result of one-shotting apps is just as bad today as it was [three years ago]

Hard to respond to this with anything other than "no, you're wrong".

What’s the production mvp app that ai made in 30 minutes then? That was the original question no?

When did you last try this? Here's a prompt for you:

> It's really important to me you make sensible decisions here, and don't bother me with the small stuff. I want a plant-watering app me and my wife can share, that shows who watered which plants in our house. I'll deploy this on my home server with Coolify. The app should be attractive, work both on desktop and mobile. We have a bunch of cases where we have multiples of one plant type. We'll need separate users, but don't go overboard with auth. I want to impress her, so let's lean on the side of more rather than fewer, features, but I don't really wanna run anything that won't just fit in a single container with some persistent storage. We're the only two users who'll ever actually give this a go. Visually attractive is important to me.

I'm just being honest but the parent said "production," is that your idea of what that means?

For an MVP? Sure. I would indeed be willing to deploy this to me and my wife.

I think we’ll see a lot of layoffs and then the tech industry is going to become more vertically, integrated with product, business analysts and developers all combining into one role. It sucks because there goes one of the highest being roles in America right now that actually employed a lot of people.

Ironically, I don’t think tech support is going to be fully replaced by these anytime soon. That’s one place you definitely need to have actual people talking to other people. Lawyers and doctors are gonna be legally protected too because you still need a human to sign off on all those actions though we will probably need far fewer.

Are the models all that much better? To me it seems the tooling surrounding the models got better, but the models themselves are basically interchangeable unless you're following a bunch of flawed overfitted benchmarks.

Also MVP apps are great and all, but I've seen 0 evidence of actually useful software from all this tooling, if anything all the software I'm using has just become more buggy and less reliable over time

I don't know, even if AI allows two engineers to do the work of six, companies will likely just use that efficiency to expand their scope. I think we'll see short-term layoffs and a more stratified engineering field during the transition, but the fundamental need for deep technical expertise isn't going away.

>I don't know, even if AI allows two engineers to do the work of six, companies will likely just use that efficiency to expand their scope.

Not really. It will be a cuttthroat landscape, and the scope wont matter as much anymore. First because everyone else will equally be able to throw LLMs at the scope, but also because the scope has natural limits: your market fit, customer expectations, and (for software/hw products) physical world/manufacturing limitations.

They'll want to reduce their margins.

AI usage will directly impact said margins. Moreover, for the scenario you describe, companies need to have the capability to precisely estimate the cost of a given deliverable - not something possible with current tooling + models. You're also underestimating the market trend towards vertical integration: companies are not going to be constrained by a sector or niche. They will expand to capture as much value as they can, because now their capacity to do so is partially decoupled from labor.

It will certainly be a cutthroat landscape for engineers, but companies will be building _more_ capacity, not less. In other words, the demand won't disappear for skilled technical labor, it will just move higher up the value chain.

>You're also underestimating the market trend towards vertical integration: companies are not going to be constrained by a sector or niche. They will expand to capture as much value as they can, because now their capacity to do so is partially decoupled from labor.

They will still totally be, because the capacity to do so was never coupled to labor, it was coupled to domain knowledge, client network, other players dominating the market, and so on...

> even if AI allows two engineers to do the work of six, companies will likely just use that efficiency to expand their scope.

1) they won't, they'll just cut costs

or

2) they will, but unless it's a new scope or one that can absorb growth, they'll just be competing with other companies in the space and taking away business from them

either way, labor loses

No, my life experience tells me those companies will fire the ones they no longer need instead.