How to play: Some comments in this thread were written by AI. Read through and click flag as AI on any comment you think is fake. When you're done, hit reveal at the bottom to see your score.got it
A general question for HN: Are there any translation services that I can recommend that people use instead just asking a standard AI to translate in the naive way? I'm saddened by the number of times I'm seeing non-native speakers get tarred with using an AI because the naive way an LLM translates slathers its own style on top of the translation rather than being a more faithful translation.
Standard translation services lack the understanding of LLMs and LLMs are hilariously prone to these AI tells, even when you ask to avoid them.
I try to learn writing and copywriting directly in english to avoid the awkward translation effect, but I focus more on the rythm, structure and content than the prasing itself
i believe that the big ones, bing, google, even deepl are all using AI to translate. the variation and the mistranslations i sometimes get are to wild.
i wish the translation tool would tell me if it can't directly translate a word because of misspellings or abbreviations. i learned that i have to write in a very explicit straight forward way, with each sentence standing on its own to make sure i get sensible translations.
> because the naive way an LLM translates slathers its own style on top of the translation rather than being a more faithful translation.
If you're sincere then I'm afraid you're being led on by the lazy.
When tasked with translating a document, unless instructed otherwise, the LLM does not inject its own cliched stories, narrative structures and assorted engagement tricks.
Of course not. What it does seem to do is inject its grammar, style, and propensity for certain constructs that cause people to flag it as "AI" in their brain.
It's possible they're all lying. It's very unlikely that zero of them are lying, as I happen to be in a place where I get exposed to this, so I've seen it a lot. But I've seen it an awful lot for it all to be liars.
Unfortunately, I don't speak a foreign language fluently enough to test it myself. I have some I can kind of verify that the translation is literally accurate but I don't have any second language I could do a style translation for.
> I built a SaaS in a few hours. It does its job better than tools people pay real money for.
It's easy to build something for yourself that solves your own use case. You know how you work and what's important to you. Most of the effort in software goes into understand people other than you and building guardrails for them.
> Whatever niche tool you’re sketching this week, a thousand people can prompt a working version by Sunday night
A lot of people on HN are naturally builders. For them it's fun to play around with AI and see what can be built. Not everyone is like that.
For individuals, how many people actually want to spend their weekends building a niche tool for themselves vs just paying a few dollars a month to access something that does the job even though it may not be perfect?
For businesses, the buy vs. build tradeoff does now have some new compelling evidence in favour of "buy" but there's still maintenance needed for the service internally, there will be people with feature requests for it. Before you know it, the company is diverting resources away from their core business for a support tool that doesn't really drive impact.
I think we'll see more options in the marketplace as individuals will be able to get started more easily but SaaS is far from dead.
> Most of the effort in software goes into understand people other than you and building guardrails for them.
100%, which is why I think hyper-tailored, small software is an exciting option. When you shift your goal to serve 1000 customers, the hardest part is not having a tech stack that supports them, but making sure the software is good enough for all their different needs. Or you try to meet them all and you get into feature creep.
> For businesses, the buy vs. build tradeoff does now have some new compelling evidence in favour of "buy" but there's still maintenance needed for the service internally
What if you neither buy or build the software but buy the service outcome?
A data analyst that creates custom workflows with auto-generated dashboards. A dev-ops that provides access with exactly the information you need, depending on the hand-off level you want for the process.
My point is that a lot of these micro tools that started to appear the last few years will vanish, either bundled with bigger SAAS, internalized, or shifted to a service offering. This trend started with productized services, this is the next step.
> SaaS is far from dead.
Micro SAAS*. I agree with all your points, but I do believe the small software landscape will change drastically in the coming years
This is just the old argument in favour of SaaS, though. The article is pointing out that those long-standing rules may no longer make sense, in a world where 1. it no longer takes even a weekend to build the niche, perfectly custom tool and 2. the maintenance / feature requests side is also just a prompt away.
At my job (we're somewhere in the transition between startup -> enterprise), we still use a lot of SaaS and are continuing to do so, but even there we're starting to move off certain platforms (Pendo, is a recent casualty) as we've realized in many cases we can just build it ourselves, pay drastically less, and actually get to control how it all works.
My suspicion, though I could be wrong: SaaS will move more and more towards focusing solely on the enterprise. Personal users and startups will have a harder time justifying both the cost and lack of flexibility of pure software SaaS. Enterprise will continue to be ok with paying the costs in exchange for compliance and vendor stuff and integration and whatever else they care about.
Even if maintainance is just a Claude/ChatGPT session away we know that the cost of that session could be 1-2 hours of attention going to a supporting tool. And what if suddently is down? now your software team is having an incident about an in house tool instead if an incident about your paying customers.
Moreso while a custom build software just for you is a great fit for your own processes you are loosing innovation and what you could benefit from using a software where someone is trying to innovate and bring out better things that you might not know even to ask for in a prompt as that is not your own domain of expertise or better said you are too expert in your own process that you cannot see what else is out there. And to overcome this you have to try to put in resources for this tool.
> those long-standing rules may no longer make sense, in a world where 1. it no longer takes even a weekend to build the niche, perfectly custom tool and 2. the maintenance / feature requests side is also just a prompt away.
…as are your own, perfectly custom security vulnerabilities and their successful exploitations.
For businesses that build software (which is more than you'd think, as this includes the entire financial sector), building vs. buying has an entirely new scale.
At work, which I won't say where, we replaced a few SaaS we were using with in-house software. We largely have a culture of in-house software so it made sense. These were not small tools. They were behemoths created by multi-billion dollar software houses. And we replaced them, entirely, in the span of weeks. The reason it worked is we already had many domain experts in those tools.
The advantages don't stop at "money". The biggest advantage is complete control. We can integrate across our other tools, for free. We can add whatever features we want.
> It's easy to build something for yourself that solves your own use case. You know how you work and what's important to you. Most of the effort in software goes into understand people other than you and building guardrails for them.
Fully agreed. It's also just a time saver. People are lazy, and I don't mean that in a bad way, just that people don't really want to sit there and explain to an agent exactly what they want software to function like. In fact, most people don't even actually know what they want software to function like.
It doesn't just stop at humans either. When I recently started creating my smart budgeting app, Claude pointed me to use Clerk for auth. So not even the AI wants to build everything from scratch.
To be fair to Claude on the Clerk thing, I had my agent build session management and auth from scratch, then beefed it up with all the best practice advice, and and and.
It took me an extra day of agent-prodding and code review.
And I still will have to add SSO and Google/whoever sign-in and passkeys and 2FA at some point.
Using Clerk would've avoided that, but at a permanent cost of one more dependency.
Before the industrial revolution or so somebody coming up in a craft would build a chunk of their toolset as part of that process. Even today many good (not necessarily software) engineers build some of their own tools. AI has made it much easier to build tooling to make your life better. Maybe some could be sold. But that isn't the intent and, as others have noted, can take significant additional effort.
I have my own "bandsintown" for example because I want to avoid the advertisement shitshow. I have a number of very specific tools for building software. I doubt I ever package up any of this for use by others much less try to monetize it.
That is the vibe I got from this article. This is a good concept to have in your (meta) toolbelt.
Contrast checkers score against solid background — did you test with actual gradient/texture behind text, not just flat swatch? APCA and WCAG both assume static bg color, real page might differ enough to matter.
Thus demonstrating that the final accessibility gate must be "the mark I human eyeball". Looks impossible to read without reader mode on my screen.
Your "paper" choice is #bfbfc4 - even going to #ddd makes the APCA go from ~63 to 80 and would probably address the complaints handily. WCAG has a lot of gaps.
Idk if I disagree completely but it feels like I do. To elaborate, I don’t think we will see less applications, I think we will see more with better features.
My reasoning is people aren’t happy with many of the titans we have for certain apps, uber is my favorite example. Everyone hates the price it charges thinks it’s unfair, but to go back to a time before uber is unfathomable.
That to me is a ripe market to be overtaken. What it really needs is a taxi driver or operator so fed up with the current system they spend their energy organizing and making their own uber with less greed.
For sure not every tool will become huge, but most of the tools we use nowadays started with someone having a problem, being frustrated with the status quo, and having the will and means to execute.
The means necessary to execute get lower and lower everyday.
Uber is not a saas, it sell the ability to match drivers with people. You can code an uber-like software easily and you will find that it leads to nowhere: no network => no driver + no consumer = you have nothing
Another kind of business is online bank: are they just software ? Yes, obviously. Are they software-as-a-service ? Obviously not: they sell regulation, integration with financial stuff, government approbation and what-not
Just because a company is used via software-only does not mean that it has a saas
On the other hand, you have companies like tesla, which are far more "software-as-a-service" than online banks
Some contrudictions in the article. Yes, Todo apps were always cheap to build, but some became large companies. Yes, the ai made the easy part easier. So why should the landscape change? Yes, technical people can build tools for themselves. Do they all really want to? What about non-technical people? Also, note - it's tools, not products. Tools help me achieve a specific goal. Products help me work better in a way I've not necessarily thought about myself.
I do think one claim is spot on - as production costs go down customization becomes cheap. This makes it economically viable to build products, especially more complex ones, for niche markets. Hubspot for medical? No problem. MailChimp for music clubs? You got it.
Ultimately, the value remained in the same place - create something people are willing to pay for. The economics might have changed, certainly for software development job market - but honestly, as a developer myself, I've always felt coding is not hard enough or valuable enough for the society to justify the insanely high salaries.
> Todo apps were always cheap to build, but some became large companies
The point is that simple apps like these have just become insanely competitive to distribute, and it's best not to think of it as mini businesses
> Tools help me achieve a specific goal. Products help me work better in a way I've not necessarily thought about myself.
For me a product = a packaged set of tools.
What was interesting with the micro-saas wave is that a lot of people started to build products around one simple tool. It worked because sometimes a well-crafted one is better unbundled. Here's one example that thrived on this premise: https://www.bannerbear.com/
I don't know if it's a coincidence but this is the same thing an AI explained to me when prompted something like "what replaces SaaS now that AI can build most things on demand". That was like a year ago. I have no proof for this, but we might all just be getting similar responses from similar contextual bubbles and landing on the same conclusions.
>> Because AI is crushing small standalone software
The only thing AI is crushing is newcomers.
Standalone software has expertise and distribution, the only 2 things that matter and that your new software does not have.
Also the needs of a business have nothing to do with software you vibe code for yourself. Its compliance, backups, reliability. Not a few features claude did for you.
I expect full circle on this one. Short term - many enthusiasts build stuff. I can fully imagine someone in a given company creating something over the weekend and persuading manager to adopt the system. However, 6 months pass, the person now moved into new role or left a company, nobody knows the product and it becomes just another SaaS just without working support behind it. So for short term it is definitely bring some losses. Long term I think it will bite those who dropped the software. Over-customisation for the needs of everyone is probably here to stay in my opinion
if it was built with AI then the knowledge of the builder is not really adding much. the next person coming in to fix issues is just going to feed the thing into an AI too, and ask the AI to fix the problem they have.
While it can be true for technical state of the system, you dont have the business decision log, nor you have the intention, and most importantly - you need someone to assume responsibility for this mini system. While it can be easier for tech companies, but assuming non tech person (say customer success manager) creates something and is replaced by other none tech person - there is significant probability that new person is simply not interested in developing system that is not part of the role. Or if it is you just limited your set of potential candidates.
Disagree that it's mostly a long-term problem. Bus-factor-one internal tools have always died quietly, AI or not. Difference now is speed of creation outpaces speed of documentation, so the abandonment cycle just compresses from years to months.
I kind of get what the article is saying. But I am looking at it from a different perspective.
The same set of things like skills, and custom tools etc can actually be used to get faster to the a product that kind of fits 1000 customers. But the change is that one can build even more custom things for some customers with those skills and tools.
Last year I built a custom integration for a customer in 2 weeks. Now I can probably do it in 2 days. What does that mean for the future, I have no idea, but surely getting custom solutions like this would be more of a moat not less.
This feels a bit like motivated reasoning to me. This little tool of yours could instead be a seed you build around. Do you have access to a market? Yes? You are in a better position than ten thousand other vibe coders with a similar idea. No? If you have the stamina to pursue a market, you have more stamina than thousands of vibe coders who will just give up.
Another way to say the same thing is that many (not all, of course) SAAS were already service with a software, especially when starting out.
Possibly.. I admit I'm quite bad with distribution, hence why I usually work with a few clients long-term.
But we do see a sharp rise in the release of apps, without the same tendency in use. [1]
I do think there will be a place for SAAS, even micro-SAAS, but the competition will be fierce and I don't have a lot of hope for indiehacker without any "unfair advantages" like unique knowledge, data, network, etc
Three days ago, Rob Walling (SaaS investor) refuted pretty convincingly the claim that genai kills SaaS. His life mission and wealth are bound up in SaaS. If he's not scared, nobody should be.
Fair that motivated reasoning could be at play. Rob's audience is SaaS founders, but whether his listeners remain engaged and his sales funnel full does not influence how genai affects their businesses, and being duplicitous doesn't fit with his character.
Adding to the topic rather than dog piling onto the website critique:
I think "the context" the author defines also includes harder to unlock peripheral components that makes software products in 2026 defensible.
For example, security is getting harder not easier. Enterprise trust rides on needing to have up-to-date knowledge on data regulation as well.
Relationships, brand maturity, discovery; all of that becomes rarified as we continue to reduce the actual product launch curve -- but also the "harder to do" hygiene stuff that takes you from a "I vibe coded this in two days, good luck" to a "your successful company can count on us" type of hardening has become progressively more of a baseline requirement.
In a way, this new dynamic of 'everyone can code' reminds me of the Unbundling -> Bundling -> Unbundling cycle of how entertainment media went from 3 broadcast channels to cable channels to web videos to our current state of bundled streaming services, with no end in sight.
Not to forget DVD shipping eras (Sony should not discontinue physical game sales, someone needs to get us silicon/holographic/terabyte recording media already), the focus is on the bundling/unbundling of software making. Slack + Salesforce vs 'we vibed our own chat and CRM just for us'...aka a Neo-SAP ERP program.
It's a cambrian explosion of software where lots are going to learn in realtime that the code was never the value of a software engineer / software, the value is in the shared understanding of an information system or workflow or business value stream, that parts of which happen to run through digital information systems.
By the time AI is done with us, its market effects mean in order to stay employable/added value beyond just widget-watchers, we will all be business strategy experts, or we will be algorithm-fed-doped-up mindless beasts, Morlocks vs Eloi, not because of AI itself, but because of the current rapaciousness of business leaders and an assumed trajectory where society doesn't adapt to social movements and actions now made possible by the existence of LLMs.
Sidneote: the LLMs themselves are not IMO going to achieve AGI any time soon. There might be a compilation of disparate systems and LLMs form part of 'tissues' of an eventual AGI, but not currently.
We will either be experts in asbestos chemistry, safety mechanisms, and remediation... or we'll be helping people tweak it for sale as kitchen sponges and pinatas.
Definitely. I wonder if this will be an increasing factor when choosing vendors. Will the security risk drive companies away from small software, due to their use of AI? Will we see a bundling of features in big established vendors?
Will they invest more into in-house security? (Doubting the last one).
For small apps used to deliver a service in my examples, I believe the risk surface is way narrower, which might make it more acceptable. For better of for worse
I don't buy it, this is essentially vibe-consulting.
Deal with one customer can be quite a nightmare, and honestly, sometimes using off the shelf solutions can result in saving time and money.
Even today I'm creating SaaS, not every problem is simple to solve. Take a look at CRMs, there are thousands of CRMs due to enormous egos wanting to build "my way", "the way", "the superior way", "our company is not like others (LOL)", etc. At the end of the day it's just a database with CRUDish UI.
My customers pay me to solve their problem well because they can not solve it themselves, when they do they just don't have the time.
Reality check: I have been coding many products with all frontiers models: Still takes a lot of time and, surprise surprise!, money for tokens. Ironically software development became more expensive for serious software that won't vibe-break/vibe-delete-prod/vibe-delete-database.
Nitpick: "micro-SaaS" already implies small/solo, so "service with software" isn't really a new category, more a rebrand of consulting-plus-tooling. Still, the underlying point stands, the code was never the moat, the ongoing labor is.
Unrelated, but since I see that the author posted this to HN: for a reader like me who likes to select text / fidget with mouse clicks as I read, this website is a nightmare.
I don't know, I feel like my OCD clicking has finally been rewarded appropriately. Usually it just results in navigating to weird places via links that are hidden in the formatting. But this time it was delightful! And I was rewarded with a game!
Reminds me of variable-ratio reinforcement schedules from operant conditioning research - unpredictable rewards tend to produce more persistent behavior than fixed ones, at least in the Skinnerian literature. Not sure the author intended a dopamine hook, but embedding one in a blog layout is a neat trick.
Yeah did similar on old blog, just checked e.getModifierState or click duration under ~150ms before firing effect, and bailed early if window.getSelection().toString() wasn't empty. Fixed 90% of complaints, cheap to add.
Oh, haha! I actually love shit like this, but ya, not in a blog as I'm also a mouse-fidgetter. You could tone activating it down a bit by only activating it on a quick click and not at all if there is selected text [/END advice-no-one-asked-me-for]
Saw this movie in the 90s with Access and FileMaker. Every shop had some guy's custom database running the warehouse. SaaS won because that guy left and nobody could maintain his mess. AI just means the guy leaves faster now.
