For one thing, the threat model assumes customers can build their own tools. Our end users can't. Their current "system" is Excel. The big enterprises that employ them have thousands of devs, but two of them explicitly cloned our product and tried to poach their own users onto it. One gave up. The other's users tell us it's crap. We've lost zero paying subscribers to free internal alternatives.

I believe that agents are a multiplier on existing velocity, not an equalizer. We use agents heavily and ship faster than ever. We get a lot of feedback from users as to what the internal tech teams are shipping and based on this there's little evidence of any increase in velocity from them.

The bottleneck is still knowing what to build, not building. A lot of the value in our product is in decisions users don't even know we made for them. Domain expertise + tight feedback loop with users can't be replicated by an internal developer in an afternoon.

The worry is that customers who do not realize the full depth of the problem will implement their own app using AI. But that happens today, too: people use spreadsheets to manage their electronic parts (please don't) and BOMs (bills of materials). The spreadsheet is my biggest competitor.

I've been designing and building the software for 10 years now and most of the difficulty and complexity is not in the code. Coding is the last part, and the easiest one. The real value is in understanding the world (the processes involved) and modeling it in a way that cuts a good compromise between ease of use and complexity.

Sadly, as I found out, once you spend a lot of time thinking and come up with a model, copycats will clone that (as well as they can, but superficially it will look similar).

The fallacy here is believing we already had all the software we were going to use and that AI is now eliminating 90% of the work of creating that. The reality is inverted, we only had a fraction of the software that is now becoming possible and we'll be busy using our new AI tools to create absolutely massive amounts of it over the next years. The ambition level got raised quite a bit recently and that is starting to generate work that can only be done with the support of AI (or an absolutely massive old school development budget).

It's going to require different skills and probably involve a lot more domain experts picking up easy to use AI tools to do things themselves that they previously would have needed specialized programmers for. You get to skip that partially. But you still need to know what you are doing before you can ask for sensible things to get done. Especially when things are mission critical, you kind of want to know stuff works properly and that there's no million $ mistakes lurking anywhere.

Our typical customers would need help with all of that. The amount of times I've had to deal with a customer that had vibe coded anything by themselves remains zero. Just not a thing in the industry. Most of them are still juggling spreadsheets and ERP systems.

AI-generated code still requires software engineers to build, test, debug, deploy, secure, monitor, be on-call, handle incidents, and so on. That's very expensive. It is much cheaper to pay a small monthly fee to a SaaS company.

I’m pretty certain AI quadruples my output at least and facilitates fixing, improving and upgrading poor quality inherited software much better than in the past. Why pay for SaaS when you can build something “good enough” in a week or two? You also get exactly what you want rather than some £300k per year CRM that will double or treble in price and never quite be what you wanted.

- anything that requires very high uptime

-very high volume systems and data lakes

-software with significant network effects

-companies that have proprietary datasets

-regulation and compliance is still very important

Then it dawned on me how many companies are deeply integrating Copilot into their everyday workflows. It's the perfect Trojan Horse.

Looks like we're headed back to the internal IT days of building customized LoB apps.

It wasn't even a coherent grammatically correct sentence that I entered and it busily went to work building the site I had in mind.

It's not hard to imagine that in the next 2-3 years, anyone will be able to build personalized apps on request.

The only named product was Retool.

Spreadsheets! They are everywhere. In fact, they are so abundant these days that that many are spawned for a quick job and immediately discarded. In fact, the cost of having these spreadsheets is practically zero so in many cases one may find themselves having hundreds if not thousands of them sitting around with no indication to ever being deleted. Spreadsheets are also personal and annoying especially when forced upon you (since you did not make it yourself). Spreadsheets are also programming for non-programmers.

These new vibe-coded tools are essentially the new spreadsheets. They are useful,... for 5 minutes. They are also easily forgettable. They are also personal (for the person who made them) and hated (by everyone else). I have no doubt in my mind that organisation will start using more and more of these new types of software to automate repetitive tasks, improve existing processes and so on but ultimately, apart from perhaps just a few, none will replace existing, purpose-built systems.

Ultimately you can make your own pretty dashboard that nobody else will see or use because when the cost of production is so low your users will want to create their own version because they would think they could do better.

After all, how hard is to prompt harder then the previous person?

Also, do you really think that SaaS companies are not deploying AI themselves? It is practically an arms race: the non-expert plus some AI vs 10 specialist developers plus their AIs doing this all day long.

Who is going to have the upper-hand?

Going back to the beginning, I think we just lack good tools for other cases where agents could be used. Copilot is not great, chatgpt alone lacks some features to use it for business as is. I think what will happen is that we will see a lot more new tooling pop up that relies on agents in niche markets which will just amplify the power users. It will be another category of SAAS the companies will adopt.

Then this project lets you generate static sites from svelte components (matches protobuf structures) and markdown (documentation) and global template variables: https://github.com/accretional/statue

A lot of the SaaS ecosystem actually has rather simple domain logic and oftentimes doesn't even model data very well, or at least not in a way that matches their clients/users mental models or application logic. A lot of the value is in integrations, or the data/scaling, or the marketing and developer experience, or some kind of expertise in actually properly providing a simple interface to a complex solution.

So why not just create a compact universal representation of that? Because it's not so big a leap to go beyond eating SaaS to eating integrations, migration costs/bad moats, and the marketing/documentation/wrapper.

But that does leave a weird gap where SaaSes that took a lot of time to make but can now be handled by an Ai won't survive either. If the business stays hand-coded it costs too much to be viable, if it moves to Ai it looses any advantage over doing it youself.

What Iam seeing is that customers are delaying purchases of large expensive software. Prime example; SAP. ECC migrations to SaaS model RISE/GROW-PublicCloud are stalling, same with onprem S4 to RISE. I see a whole bunch of my customers instead go with retaining the core but modernize surround apps with intelligent custom apps without feature bloat. For now, SAP/oracle/whatever remains the system of record, the edges are going away. I guess the same is likely happening in other spaces.

This change is coming. Definitely. The current moats around SaaS will fall and the alternate ecosystem might not have moats at all.

You do not want to have to plan out a system with 800 unique requirements yourself. It takes a ridiculous amount of work, and you are then stuck maintaining it.

It’s never been about the difficulty of programming, it’s been about the pain of designing and maintaining.

Agents offer a very attractive level of abstraction, and its customers, aren't necessarily human: it could be other agents. Many saas we have today would simply be unnecessary in the future.

If 90% of the actual work is waiting for Agent to get the work done, why would those companies keep paying those saas companies license fee per seat? It doesn't make sense.

The problem is, nobody knows how much and how fast AI will improve or how much it will cost if it does.

That uncertainty alone is very problematic and I think is being underestimated in terms of its impact on everything it can potentially touch.

For now though, I've seen a wall form in benchmarks like swe-rebench and swebench pro. Greenfield is expanding, but maintenance is still a problem.

I think AI needs to get much better at maintenance before serious companies can choose build over buy for anything but the most trivial apps.

A couple of them mentioned that they plan to cancel subscriptions totaling more than $100k/year for the apps they will replace with that SaaS. According to them, they have many subscriptions they keep only because of one feature. Another issue is that their workflows become a real mess when they need to copy and paste data into multiple tabs. Custom-built internal tools seem like an obvious solution. Those who migrate to custom-built tools, however, will face the challenge of orchestrating their lifecycle and creating a consistent deployment workflow, but this is one of the challenges we are trying to solve at UI Bakery.

In my understanding, SaaS products that provide customers access to proprietary data are in a much better position than other SaaS platforms. HubSpot’s acquisition of Clearbit a couple of years ago now makes even more sense because it will help them retain some of their clients.

As for Retool, I see the several waves of low/no-code products, the current one being LLMs, as repeated attempts to get non technical idea-guys to build their ideas. Where they all fail, and this is fundamental to the problem they're trying to solve, is that idea-guys' ideas crack when meeting reality. And neither Retool nor LLM fix that.

When it comes to SaaS that's industry specific, I just don't see it'll be that much of a change any time soon. I've worked heavily in the engineering industry and the security requirements that get put upon anything are nuts. It is difficult to enter this market, ISO compliance is important, even being in the cloud is a barrier for some customers, and often the type that you have no choice but to contract with if you want to make a profit because of their outsized importance in the market.

When I speak to customers, they actually quite often have tried to build something themselves. Usually it's been an intern or grad trying to make their life easier. Often it's spreadsheet based, but some go as far as knocking up little Python web apps. In one company I interned in they had a shadow PHP app. They often have a small 'data science' team that has struggled to get access to the data they need. While they can often get something that does the barebones of the tasks, and can do it well, where they fall down is that they're vulnerable to security issues and can't navigate their internal company politics to get permission to host things in the cloud and make their life easy, plus they don't have the experience to know what's good practice. I don't see AI changing things that much in that.

This practice predates even SaaS.

I read this article expecting to see a specific SaaS that was at risk, and the most I saw was "dashboards." (Which: dashboards frequently aggregate data, while the ongoing work of collection/maintenance/etc. is done by more complex applications.)

The thesis seems to be that companies can use coding agents to build one-off internal versions of SaaS apps like e.g. Workday or Salesforce or Slack or Jira or MixPanel or HubSpot. Which, if one could make such a thing for free and maintain it for free, why not?

Fortunately/unfortunately depending on where you sit, magical thinking isn't going to get Claude Code to build Workday, regardless of the quality of your AGENTS.md. Sometimes I wonder if the people who write these takes have spent any real time using Claude Code. It's good, but please be realistic.

This is inevitable, you can't rely on user licenses as a growth metric

The optimistic angle nobody's exploring: maybe 'eating SaaS' means we finally escape the subscription hellscape where every basic function costs $29/month. If an AI agent can stitch together free/cheap APIs instead of forcing you into Notion/Airtable/Whatever, that's not destruction—that's evolution.

Our customers ask for about AI features and it’s a constant struggle to explain to them that they just aren’t there yet.

With AI, that equation is now changing. I anticipate that within 5 years autonomous coding agents will be able to rapidly and cheaply clone almost any existing software, while also providing hosting, operations, and support, all for a small fraction of the cost.

This will inevitably destroy many existing businesses. In order to survive, businesses will require strong network effects (e.g. marketplaces) or extremely deep data/compute moats. There will also be many new opportunities created by the very low cost of software. What could you build if it were possible to create software 1000x faster and cheaper?"

Paul Bucheit

https://x.com/paultoo/status/1999245292294803914

I think this sort of ignores the fact that S&M agentic tools exist and the cost of those services is also dramatically decreasing, so does it net out and just become a more efficient model in general?

I'm not a consultant anymore but my friend who owns a dental clinic asked me if I could build them a personalized system that checks in with the staff every week; a thing that helps analyze how they feel week to week and helps my friend update her management strategy and coaches her on how to talk with her staff / helps her figure out her staff's communication strategies and what work they prefer to do; and she'd like me to run and host it so she can't see the raw data from her staff so they'll trust it more as it's run by a third party.

She could probably figure out how to do this but she'd still rather pay me like $5k to do this than spend 100+ hours figuring this out herself. Even with AI it'd probably take me at least a couple of weeks to get it working 100% as intended, and I don't have a dentist business to run.

I think we'll see more back office SaaS, becuase the problems to solve are near infinite, and no one has time to build all these themselves.

- modest incremental gains in productivity

- society will remain mostly the same

- very few people will take advantage of the opportunities unlocked by AI

https://efitz-thoughts.blogspot.com/2025/08/the-effect-of-ll...

But who's going to deploy it, make backups, integrate authentication, review security?..

Now, perhaps, it would be nice to have some kind of a ERP framework which would host AI-generated apps and connect them to each other. Is there anything like that?

A tangent, I feel, again, unfortunately, the AI is going to divide society into people who can use the most powerful tools of AI vs those who will be only be using chatGPT at most (if at all).

I don't know why I keep worrying about these things. Is it pointless?

1) helping to saturate traditional SaaS because code is being commoditized / the effort to build is dropping significantly.

2) defining an adjacent sub-category of SaaS: "Service-as-a-Software" where the SaaS provides _outcomes_ instead of _tools_; this couldn't really exist at scale before recently.

Automation is not new. What's new is the capabilities of the models that can be assigned with some of the workflow steps. If these steps were served by SaaS companies so far, they will still serve it. Maybe they make it much cheaper and use a model themselves.

Secondly, the way this person describes "agents" is not rooted in reality:

>Agents don't leave. And with a well thought through AGENTS.md file, they can explain the codebase to anyone in the future.

>What's going to be difficult to predict is how quickly agents can move up the value chain. I'm assuming that agents can't manage complex database clusters - but I'm not sure that's going to be the case for much longer.

What in the world. And of course he's selling a course. This is the same business as those people sitting in Dubai selling $6000 options trading courses to anyone who believes their bullshit. The grifter economy around AI is in full swing like it was around blockchain/crypto in 2017-2020.

They are also on the basis of high gross margins of 80-90%. What happens to margins when you start including token variable costs?

The first company was a low margin business that sent home health care nurses to special needs kids and reimbursements came from Medicaid.

I was hired by the new director to modernize their aging in house Electronic Medical System built on FoxPro 1999 running on SQL Server 2000 - in 2016.

They had two “developers” who had been their for 10 and 20 years respectively who only knew Sql Server and FoxPro.

They also had some other software.

After doing some assessments of the situation, my report to the director and the CTO was that this company should not try to support a software development department and hire new people. Their margins are too small to be competitive or to keep people.

I suggested we outsource everything to other consulting companies - not staff augmentation. Let the consulting company do the entire implementation based on a Statement of Work.

The two “developers” role changed to “data analyst”. Even with AI I would have said the same thing today. Not every company needs to try to do software engineering. Every company does need to understand its data. [1]

The next company was a startup. I was adamant about blocking every developer who suggested any internal tool that we could get a well known SaaS to do or where AWS had a service that wasn’t firefly related to our product. To use the cliche - anything “that didn’t make the beer taste better”. My opinion wouldn’t have changed with AI.

The last thing I want is a bunch of bespoke internal vibe coded AI Slop that we have to support that is not in service to the product when we can find a reputable third party product.

And no that doesn’t mean I am going to trust some unknown one person SaaS company.

[1] 18 months into the job, I walked into the director’s office and told him, “let’s be honest, you all don’t need me anymore”. I purposefully put myself out of job. But boy did I have a story to tell during behavioral interviews at my next job at the startup and my interview for my job at BigTech after I left the startup.

At the same time, to the core theme of the article - do any of us think a small sassy SaaS like Bingo card creator could take off now? :-)

https://training.kalzumeus.com/newsletters/archive/selling_s...

I think that _developers_ might be reaching for more LLM-built tools instead of SaaS in some cases and I also can believe that plenty of people _think_ they are vibe-coding up alternatives to SaaSes they pay for but I think those people are going to have a bad time when it eventually collapses (the tool they made, not talking about the AI bubble).

I'm not anti-LLM (not in the slightest) and you can sometimes (it's not a given) get to 80-90% of an existing product/service with vibe-coding or LLM-assisted development but that last 10-20% (and especially that last 1-5%) are where it gets hard. Really hard.

It's the typical "you can already build such a system yourself quite trivially"-mentality IMHO. I feel this myself all the time, even before LLMs, "Oh, I could clone this easily!" and in many cases I could or even did... or at least I cloned the easy/basic/happy-path version that eschewed a whole slew of features I didn't need/care for. But then the complexity started to set in. [0]

I have the same feeling for things I'm not even trying to clone, just build from scratch. I put together a cookbook for friends and family recently and used LLMs to help write essentially a static site generator to read my JSON data I created (some with the help of LLMs) and render it out as HTML (which I then turned into a PDF). My mind started to run with "Hmm, could I create a product out of this? It was relatively easy to get started..." but then reality set in and I remembered all the little tweaks I had to do (shorten a title here, reduce padding there, etc to make everything fit and look good). Sure, I got 80% of the way there in the first or second iteration of it using LLMs but there was plenty of massaging that had to happen to turn it into something usable that I could send to a printer.

All you've done is swapped a SaaS built for your problem domain with another, more expensive SaaS that has no support at all for your actual problem. Why would anyone want that? People buy SaaS products because they don't want to solve the problem, they just want it fixed. AI changes nothing about that.

Oh, child.... building is easy. Coordinating maintenance of the tool across a non-technical team is hell.

> But my key takeaway would be that if your product is just a SQL wrapper on a billing system, you now have thousands of competitors: engineers at your customers with a spare Friday afternoon with an agent.

I think it’s pertinent to point out that a lot of SaaS products are aimed at businesses and individuals who don’t have engineers at all.

AI agents aren’t going to disrupt the SaaS market for software intended for businesses like small business retail where the owners and staff have minimal technical knowledge and zero extra time.

I also think that some SaaS products are so cheap that about an hour of effort is too much. Is it worth a month of effort to vibecode a Dropbox alternative? Even some pretty complicated software that is untouchable by agents and engineers’ side projects like the Microsoft 365 suite and Jira are priced at under $20/month/user.

On the other hand, some entrenched solutions that aren’t all that complicated could be finding themselves with new, smaller competitors.

SaaS are swiss-army knife tools and you don't need all of this.

do you want to have a contact form on your site? Don't but the whole WP plugin for forms, ask AI for tiny, well-aligned plugin which will display form fields and process the input.

Do you need to A/B test your landing page? Just ask for another plugin which will switch page versions and track impressions.

No need for Hubspot when you have google sheets + AI-made plugin for this.

Summary is that for agents to work well they need clear vision into all things, and putting the data behind a gui or not well maintained CLI is a hinderance. Combined with how structured crud apps are an how the agents can for sure write good crud apps, no reason to not have your own. Wins all around with not paying for it, having a better understanding of processes, and letting agents handle workflows.

SaaS maintenance isn't about upgrading packages, it's about accountability and a point of contact when something breaks along with SLAs and contractual obligations. It isn't because building a kanban board app is hard. Someone else deals with provisioning, alerts, compliance, etc. and they are a real human who cannot hallucinate that the issue has been fixed when it hasn't. Depending on the contract and how it is breached, you can potentially take them to court and sue them to recover money lost as a result of their malpractice. None of that applies to a neural network that misreads the alert, does something completely wrong, then concludes the issue is fixed the way the latest models constantly do when I use them.

1. I had two text documents containing plain text to compare. One with minor edits (done by AI).

2. I wanted to see what AI changed in my text.

3. I tried the usual diff tools. They diffed line by line and result was terrible. I searched google for "text comparison tool but not line-based"

4. As second search result it found me https://www.diffchecker.com/ (It's a SaaS, right?)

5. Initially it did equally bad job but I noticed it had a switch "Real-time diff" which did exactly what I wanted.

6. I got curious what is this algorithm. So I asked Gemini with "Deep Research" mode: "The website https://www.diffchecker.com/ uses a diff algorithm they call real-time diff. It works really good for reformatted and corrected text documents. I'd like to know what is this algorithm and if there's any other software, preferably open-source that uses it."

7. As a first suggestion it listed diff-match-patch from Google. It had Python package.

8. I started Antigravity in a new folder, ran uv init. Then I prompted the following:

"Write a commandline tool that uses https://github.com/google/diff-match-patch/wiki/Language:-Py... to generate diff of two files and presents it as side by side comparison in generated html file."

[...]

"I installed the missing dependance for you. Please continue." - I noticed it doesn't use uv for installing dependencies so I interrupted and did it myself.

[...]

"This project uses uv. To run python code use

uv run python test_diff.py" - I noticed it still doesn't use uv for running the code so its testing fails.

[...]

"Semantic cleanup is important, please use it." - Things started to show up but it looked like linear diff. I noticed it had a call to semantic cleanup method commented out so I thought it might help if I push it in that direction.

[...]

"also display the complete, raw diff object below the table" - the display of the diff still didn't seem good so I got curious if it's the problem with the diffing code or the display code

[...]

"I don't see the contents of the object, just text {diffs}" - it made a silly mistake by outputting template variable instead of actual object.

[...]

"While comparing larger files 1.txt and 2.txt I notice that the diff is not very granular. Text changed just slightly but the diff looks like deleting nearly all the lines of the document, and inserting completely fresh ones. Can you force diff library to be more granular?

You seem to be doing the right thing https://github.com/google/diff-match-patch/wiki/Line-or-Word... but the outcome is not good.

Maybe there's some better matching algoritm in the library?" - it seemed that while on small tests that Antigravity made itself it worked decently but on the texts that I actually wanted to compare was still terrible although I've seen glimpses of hope because some spots were diffed more granularly. I inspected the code and it seemed to be doing character level diffing as per diff-match-patch example. While it processed this prompt I was searching for solution myself by clicking around diff-match-patch repo and demos. I found a potential solution by adjusting cleanup, but it actually solved the problem by itself by ditching the character level diffing (which I'm not sure I would have come up with at this point). Diffed object looked great but as I compared the result to https://www.diffchecker.com/ output it seemed that they did one minor thing about formatting better.

[...]

"Could you use rowspan so that rows on one side that are equivalent to multiple rows on the other side would have same height as the rows on the other side they are equivalent to?" - I felt very clumsily trying to phrase it and I wasn't sure if Antigravity will understand. But it did and executed perfectly.

I didn't have to revert a single prompt and interrupted just two times at the beginning.

After a while I added watch functionality with a single prompt:

"I'd like to add a -w (--watch) flag that will cause the program to keep running and monitor source files to diff and update the output diff file whenever they change."

[...]

So I basically went from having two very similar text files and knowing very little about diffing to knowing a bit more and having my own local tool that let's me compare texts in satisfying manner, with beautiful highlighting and formatting, that I can extend or modify however I like, that mirrors interesting part of the functionality of the best tool I found online. And all of that in the time span shorter than it took me to write this comment (at least the coding part was, I followed few wrong paths during my search for a bit).

My experience tells me that even if I could replicate what I did today (keeping motivated is an issue for me), it would most likely be multi-day project full of frustration and hunting small errors and venturing into wrong paths. Python isn't even my strongest language. Instead it was a pleasant and fun evening with occasional jaw drops and feeling so blessed that I live in SciFi times I read about as a kid (and adult).

I run Cloude Code to build that SaaS for me locally over the span of few weeks, to get the same value.

This article makes no sense.

How wrong the author is about that! IMO As soon as the bubble bursts, which is already evident and imminent, these agents will raise their subscription fees to ridiculous amounts. And when that happens, entire organizations will abandon them and return to good ol' human engineering

> The signals I'm seeing

Here are the signals:

> If I want an internal dashboard...

> If I need to re-encode videos...

> This is even more pronounced for less pure software development tasks. For example, I've had Gemini 3 produce really high quality UI/UX mockups and wireframes

> people really questioning renewal quotes from larger "enterprise" SaaS companies

Who are "people"?

This is the key point. Sure, you don't have the chops to be able to replicate the SaaS product locally with Claude/Gemini, but you don't have to, because you're no trying to make a product that can handle N+1 workflows.

Classic tell this is tech bro-ese.