From my understanding, they vibe coded a solution and for now it is working for them. I've worked with zendesk and several other platforms. I worked in customer service automation, so we had to deal with so many incompatibility with different platforms. This looks like a win for now, but the part we won't see is what happens at 72 hours.
I'm all for sticking it to zendesk, but as I tell every single person who were thinking to roll their own solution, have you thought about Integration? That's zendesk's moat. They have an integration with almost every single platform you can think of. It works with all e-commerce, but also ebay and amazon. It can communicate via WhatsApp, imessage, signal, and everything in between. Then connects to salesforce and netsuite.
I know this is a AI generated post, talking about an AI generated app. So next I'm expecting the AI agent deleted our prod database posts.
I know you had commercial arguments when working at Zendesk.
Still: when you code your own solution, you don't need to build an integration with every single platform you can think of. Just those you use.
And sometimes, especially now with AI, coding your own solution exactly tailored to your needs can be simpler than configuring a complex product designed to match as many use case as possible.
>And sometimes, especially now with AI, coding your own solution exactly tailored to your needs can be simpler than configuring a complex product designed to match as many use case as possible.
Coding a solution was never a problem. Supporting and maintaining it was. I can guarantee you an in-house ticketing system will be more expensive than Zendesk for every small and medium company.
I've replaced Zendesk and Intercom with my custom solution built with Go and Sqlite.
It's in production for 4 months now.
I've not felt need for anything. We 2-3 custom integration built into it.
The only issue I had was spam. So I thought about adding turnstile but guess what? Cloudflare turnstile was giving me some error so I just added honey pot fields and didn't have any issue so far.
And also sqlite+go is blazingly fast, our agent love the speed.
It's running on $5-6 Hetzner server. Backups using Litestream to s3.
> I know this is a AI generated post, talking about an AI generated app. So next I'm expecting the AI agent deleted our prod database posts.
This is like the new racism, just call something that you don't like or want to denigrade as "AI". It works on both sides, both AI lovers and AI haters.
If only you cared to read the article.
A human spent time writing it, they gave two purposefully narrow commitments that if someone care to read the article, you'd know comes from a human.
Since their private equity buyout in 2022, Zendesk will absolutely attempt to get you to sign agregious terms. They will attempt to hoodwink you. They will pretend they never received your cancellations and charge you for another year of contract. Legal counsel should be leveraged when dealing with them to ensure you don't get screwed.
Yeah but they're boring. Hilton, Skype, and Dell are PE success stories. PE, when they're not being egregious, makes the website go down a bit less? Support improves? Security and compliance gets a bit more spent on them and they land more enterprise deals? Hardly the stuff to counteract taking down everyone's beloved Toys'Я'us with leveraged debt.
Sure, in the case of private equity rollups in HVAC and electric and plumbing you often get higher rates but professionalized systems including real contracts, phone answering, emailed invoices, real quotes, real terms, enforceable warranties. Many who have trusted vendors may view this as degradation because of the cost increase but for others (a new homeowner with a job), the professionalization is a positive.
For software like Zendesk that was already thoroughly professionalized, I agree, it's hard to think of positive attributes to a PE buyout!
i've gone down this route before, we're back on freshdesk now. it's easy to build a prototype in two days, but the long tail of making it actually meet _all_ the requirements is hell.
the value platforms provide is that _you_ didn't make the software, somebody else did, so somebody else defines the functionality and workflow of the software. it can be treated as fixed, and a thing that people learn how to use. they have support docs and training resources for your staff to access. when somebody has a problem, you can tell them to make a ticket with freshdesk not with you.
if you make the software in-house, you have to also make all those training resources yourself. you have to make all the ui decisions. and you have to defend those ui decisions, even when each of your support staff wants something different, and knows there's nothing preventing you from changing it to exactly what they want. even when exactly what they want contradicts with what the person sitting next to them wants.
As someone who is currently in the process of integrating Zendesk: It seems to have a moat, however directed in the wrong direction. This integration might be my first project failure in the last decade which included two SAP integrations.
There is just so much clunk and developer hostile stuff going on that I would rather just not deal with it anymore.
It's that last S in SaaS that builds the moat around companies like this. Even if you can make the Software, it's on you to provide the Service as well if you go that route. Quite a lot of companies simply don't want to self-host yet another thing and make their IT group even larger (or more burdened). It's rarely the Software or the Platform that's the selling point as much as the Service.
or, as you have been able to do for decades, use an existing open source solution? Zendesk’s value isn’t in the CRUD operations. If a company doesn’t want to pay for Zendesk, don’t ask Claude to build a replacement… use a good free solution, that has years of domain expertise built into it.
If you have an internal CRM and are maintaining a Zendesk integration to create support tickets within your product, you've likely already done half the work needed to instead create those tickets in your own internal CRM tool yourself. This makes a lot of sense.
As someone who's had to maintain a Zendesk integration such as this for a large app it's hard to understate the benefit of having all that support info so close to the rest of your user's data. I have seen a huge amount of effort go into trying to get just the right balance of data in and out of ZD. Also helps alleviate concerns with sharing too much customer data with a third party.
This definitely isn't the right call for everyone, but there's a lot of upside if it can work for your organization.
Great question! This is certainly a topic that touches on many important aspects of our modern world, and there are a lot of nuanced perspectives worth considering.
it's really sad. the internet is slowly dying (along with humans using their brains, it seems). On twitter, any time anything interesting is posted (which also is usually "enhanced" with AI) 80% of the responses are all very clearly just claude/gpt replies. what is the fucking point? whats the end game for these accounts?? i HATE that i have to sit there parsing through all of this cruft. Fuck this whole article could've been so much shorter, and more concise if a human just sat there, timeboxed themselves, and wrote!
No, it's just that people who were too lazy to write anymore than a paragraph can now paste that paragraph into a chatbox and get a full blogpost. The only issue is that it still contains only as much content as was in the single paragraph, the rest filler.
Actually LLMs are like the ultimate laziness enabler. But as people adjust their expectations this stuff will no longer look impressive, since people will know it's just AI slop, and the lazies will fade back into obscurity.
Hahaha totally relatable. Love Zendesk but they kept charging and adding more features I don't need.
I built https://pointanswer.com/ for myself to host 3 of my own SaaS instead of paying more than $100/mo for simple helpcenter. I'm the only one using it with no customers but it's still way better than paying for Zendesk.
But did spent too much time on it as I built in pre-Claude era.
My personal experience:
For many years I used Zendesk to manage support and host my documentation. It's a powerful platform with help center structure that I liked liked the most. I was paying about US$30/mo for OnVoard.
When I use Zendesk for my second SaaS business, RenderKu, it costs me $70/mo. This amount is more than 3 Hetzner servers I'm paying to host the whole infrastructure. And I'm only choosing Zendesk mainly for hosting documentation.
At this point, when I was planning to start my 3rd app business, the thought of forking another $70/mo for hosting documentation made me rethink my options. I've eventually come to the conclusion it would be better for me to start PointAnswer and use it as helpcenter for myself since I only needed simple and affordable helpcenter.
Isn't this a perfect case for why SAAS stocks have been puking all year?
I certainly don't think Zendesk's core business is threatened here, and I have no desire to replace Zendesk with a custom solution. But Zendesk's ability to upcharge, sell adjacent products, and pass on cost increases is hugely threatened. If PE was planning on a decade of consistent cost increases and more land, that dream seems materially threatened!
I have a friend who’s doing this for himself. He owns an AC business. He has need for tools but does not like plunking down money for a feature set that moves him in a direction he doesn’t wish to go. Solution? Create a bespoke internal system of one off apps for his OWN business using an LLM.
He’s not a software developer, he has no concept of software maintenance or security.
I’ve been watching and it’s interesting to me because I would not be surprised that he’s not alone as a small family business. Many probably feel liberated from company’s that would enforce a certain cookie cutter shape.
Does this mean AI is shifting towards contractor jobs more? Does it mean a huge security issue brewing? Both? Maybe business owners turned SWVibers like him will swing back to an off the shelf option once pouring more effort into 3am-my-stuff-is-broke scenarios becomes more of a chore than it’s worth.
I feel like there are a million billion green field projects brewing that will soon turn brown for one reason or another.
> Maybe business owners turned SWVibers like him will swing back to an off the shelf option once pouring more effort into 3am-my-stuff-is-broke scenarios becomes more of a chore than it’s worth.
I mean, that's what, as an industry, we're all desperately hoping for, but, well, shit. ChatGPT-5.5 is quite capable, so if the business owner is disciplined and prompts it with thought, I'm not so sure those greenfield projects are going to turn brown. Hell, if I was an AC business owner, would I rather pay a software startup who think they know the AC business, or pay another AC business owner to use the software they wrote?
"Your job isn't going to be taken by AI, it's going to be taken by someone using AI" -Jenson Huang
Problem is, that person using AI is from outside our field. (Not a problem for the AC business owner who has a new product to sell! Good on them, if they choose to go that route.)
We ended up doing that ourselves, moved completely to an internal system that integrates well with our own administrative capabilities. It has been such a breath of fresh air, and took us about the same time to accomplish.
The final straw was attempting to move Zendesk down to one seat to have historical/archive capabilities for a period of time, and they couldn't even manage that process properly.
that's actually NOT what you want to do. You want to build out a clone for these SAAS product that don't have the features you don't need, eliminating footguns and accelerating the fast path to user success.
Large companies may have difficulty embracing this strategy because software is a cost center, and not a revenue center, for them. The returns to efficiency gains are really hard to measure.
Yes, I want to build a clone for the SAAS product my company uses.
And if someone has gone to the bother to make a prompt that helps my LLM build one, and it either already doesn't have the features I don't need, or I can have my LLM start by trimming the features I don't need before it builds it...
Then I think you're saying the same thing I'm saying.
It certainly is true that many SaaS subscriptions can be replaced with custom written(supercharged by AI) tools. The key question however remains “Who are you going to call when something goes wrong?” Or “who is going to maintain this software?” Because essentially that’s what SaaS subscriptions genuinely offloaded and let you focus on your core business things.
As is most things in engineering, this is a trade-off and the line where buy vs build exists somewhere to be found and justified. Sure AI moves the line closer to build these days.
> We pulled an ultrahackathon. The team who built this did not sleep much for two days. They worked through Wednesday night, through Thursday, through Thursday night, through Friday. They ate at their desks. They wrote the spec late Wednesday evening and they wrote the cutover commit on Friday afternoon, and in between they did the work that the time between those two moments required.
Cool story but I would not want to be in their shoes. Treating your employees poorly only to justify overnight changes in business needs creates a highly toxic work environment.
Yeah that was a huge red flag. Classic power tripping CEO move to arbitrarily inject a deadline for a new project mid-sprint which then forces your team to work all-nighters and still have them deliver on all their sprint work.
If I worked there I'd be immediately looking for a new employer, I guarantee this isn't the first time they've done stuff like this over the years.
Yeah. I understand that this was an exceptional situation outside their control, but having to work multiple days straight is nothing to be proud of. It's something to apologize for and give a corresponding monetary reward for, and then see what kinds of mitigations you can take to avoid this in the future.
>The honest answer is the boring one. We pulled an ultrahackathon. The team who built this did not sleep much for two days. They worked through Wednesday night, through Thursday, through Thursday night, through Friday. They ate at their desks. They wrote the spec late Wednesday evening and they wrote the cutover commit on Friday afternoon, and in between they did the work that the time between those two moments required.
Man, they must have gotten paid crazy overtime. Kudos!
Our helpdesk provider wanted a three year annual commitment at double the price we've been paying after they got bought by private equity. We are not continuing with our helpdesk provider.
The correct answer to this behavior needs to be "lol no" until these companies learn this behavior is unacceptable at any price. They could've boiled the frog at a 10% annual hike and nobody would've cared, but you double and you are fired as a vendor and that's that.
This is precisely the future of consumer software engineering (no I’m not saying anything here about building robots rockets medicine etc. at least for the next decade probably)
This idea that software developer productivity being the goal for AI companies is just not it - every piece of code you put into an LLM is a you giving these giant companies your expertise
People need to remember where the bulk of engineering money goes: consumer advertising and consumer facing applications, and don’t forget most of them have tragically bad user experiences and dark patterns because they’re trying to make the software as a service model work for investors
if I can spend a week replicating your software that has all these bullshit dark pattern features and I can replicate that for my own why why wouldn’t I?
I'm curious whether people feel like UX could at all be a differentiator here. like, sure, i can build a very quick crud app with Ai in an hour, but there are lots of UX decisions that, if not prompted in the right direction, AI just handles badly.
i guess the problem is that once a teamp goes through the process of figuring out good UX for a certain flow, which can take time, that UX then becomes trivial to copy.
Every product reveals its design if you know how to interrogate it.
So knowing a software architecture for something you use is the HARDEST part to observe and the UX is the easiest to observe.
provided you can describe what you observe, and your desired workflow that matches your need, then you can replicate it provided you understand how to test and iterate, which again is trivial to learn.
Combine:
Observations of workflows to implement and
Notional data architecture
You can create a slimmed down version of pretty much anything.
I mean this is basically every image editor compared to Photoshop.
A designer that is used to all the features Photoshop has, and then you just use the most common workflows that most people use, to be the feature bootstrap for your Photoshop alternative that’s much lighter weight and cheaper etc.
The devs doing these kinds projects no longer think of the code as maintainable, but disposable. It's quite the 180 for a community obsessed with reusable npm packages and not reinventing the wheel. Last year Cursor bragged about ditching their CMS in 3 days: https://x.com/leerob/status/1999513884382597485
> if I can spend a week replicating your software that has all these bullshit dark pattern features and I can replicate that for my own why why wouldn’t I?
Have fun Chesterton's fencing yourself into much of what other companies spent years learning and solving.
Prototypes and proofs of concept were easy even before AI.
Which is why you start from the principles that were “learned and solved” in the past.
Literally none of those things that you’re talking about Chestertons fence hopping are secrets or hidden
They’re very well documented in product design, project management, cybernetics, IT etc…
There’s been thousands and thousands of articles written about all the different ways to build: functional programming, monorepo vs micro services, strong typed, memsafe etc….
there’s infinite number of framework debates that are already embedded in the LLMs that you simply need to interrogate as part of the planning process before you ship anything
It’s actually really simple.
The reality is that 99% of software engineers never gave care about any of that ever.
Go ask your average software engineer about Parness information hiding or Conway law inversions.
These things are trivially easy to grasp and make engineering intuitively simple to avoid MOST problems. I am consistently teaching principal engineers, VP, senior software engineers, people who know how to write code about basic basic basic information theory. It’s honestly embarrassing.
This happened to me recently with a few providers. Their billing teams got aggressive and I said fuck it and AI coded a replacement in a few nights/weekends. For me it was: Temporal, Zapier and DocuSign
I strongly believe the build/buy equation is much different in 2026 than it was in 2024
Yeah, this is the absolute last thing I'd vibecode. This is one step above sending your LLM to represent you in front of a court when you're accused of murder.
My successful startups had their own internal CRM as distinguishing factor to their success, it's not hard.
One started out as an acquisition of a group on sales force, we wrote migrated to our brand new crm the first month and integrated our PBX and several other functions within the first 3 months.
It's not hard, most CRMs don't need "webscale", you can whip up a highly targeted and integrated CRM in rails in the blink of any eye.
That's why I'm pretty sure in 2001/02 I had the earliest real customer, real time, audio reviews published online in the world.
Like any decision during a startup buy/build is a situation specific decision you have to make at every point and 3rd party CRMs are fine to launch with... but, I mean, come on.
From my understanding, they vibe coded a solution and for now it is working for them. I've worked with zendesk and several other platforms. I worked in customer service automation, so we had to deal with so many incompatibility with different platforms. This looks like a win for now, but the part we won't see is what happens at 72 hours.
I'm all for sticking it to zendesk, but as I tell every single person who were thinking to roll their own solution, have you thought about Integration? That's zendesk's moat. They have an integration with almost every single platform you can think of. It works with all e-commerce, but also ebay and amazon. It can communicate via WhatsApp, imessage, signal, and everything in between. Then connects to salesforce and netsuite.
I know this is a AI generated post, talking about an AI generated app. So next I'm expecting the AI agent deleted our prod database posts.
I know you had commercial arguments when working at Zendesk.
Still: when you code your own solution, you don't need to build an integration with every single platform you can think of. Just those you use.
And sometimes, especially now with AI, coding your own solution exactly tailored to your needs can be simpler than configuring a complex product designed to match as many use case as possible.
>And sometimes, especially now with AI, coding your own solution exactly tailored to your needs can be simpler than configuring a complex product designed to match as many use case as possible.
Coding a solution was never a problem. Supporting and maintaining it was. I can guarantee you an in-house ticketing system will be more expensive than Zendesk for every small and medium company.
I've replaced Zendesk and Intercom with my custom solution built with Go and Sqlite.
It's in production for 4 months now.
I've not felt need for anything. We 2-3 custom integration built into it.
The only issue I had was spam. So I thought about adding turnstile but guess what? Cloudflare turnstile was giving me some error so I just added honey pot fields and didn't have any issue so far.
And also sqlite+go is blazingly fast, our agent love the speed.
It's running on $5-6 Hetzner server. Backups using Litestream to s3.
Would love to see what that Intercom replacement looks like! :)
> I know this is a AI generated post, talking about an AI generated app. So next I'm expecting the AI agent deleted our prod database posts.
This is like the new racism, just call something that you don't like or want to denigrade as "AI". It works on both sides, both AI lovers and AI haters.
If only you cared to read the article.
A human spent time writing it, they gave two purposefully narrow commitments that if someone care to read the article, you'd know comes from a human.
Since their private equity buyout in 2022, Zendesk will absolutely attempt to get you to sign agregious terms. They will attempt to hoodwink you. They will pretend they never received your cancellations and charge you for another year of contract. Legal counsel should be leveraged when dealing with them to ensure you don't get screwed.
> Since their private equity buyout in 2022,
Does anyone know of examples of where a private equity buyout has made things better for the consumer?
Not quite PE, but Berkshire Hathaway generally doesn't destroy businesses that it purchases.
Yeah but they're boring. Hilton, Skype, and Dell are PE success stories. PE, when they're not being egregious, makes the website go down a bit less? Support improves? Security and compliance gets a bit more spent on them and they land more enterprise deals? Hardly the stuff to counteract taking down everyone's beloved Toys'Я'us with leveraged debt.
Sure, in the case of private equity rollups in HVAC and electric and plumbing you often get higher rates but professionalized systems including real contracts, phone answering, emailed invoices, real quotes, real terms, enforceable warranties. Many who have trusted vendors may view this as degradation because of the cost increase but for others (a new homeowner with a job), the professionalization is a positive.
For software like Zendesk that was already thoroughly professionalized, I agree, it's hard to think of positive attributes to a PE buyout!
For real. Does stuff like Zen desk even have any moat at all? It's an easy framework crud app with a million features that you don't need.
Just get an AI bot to make one for you
i've gone down this route before, we're back on freshdesk now. it's easy to build a prototype in two days, but the long tail of making it actually meet _all_ the requirements is hell.
the value platforms provide is that _you_ didn't make the software, somebody else did, so somebody else defines the functionality and workflow of the software. it can be treated as fixed, and a thing that people learn how to use. they have support docs and training resources for your staff to access. when somebody has a problem, you can tell them to make a ticket with freshdesk not with you.
if you make the software in-house, you have to also make all those training resources yourself. you have to make all the ui decisions. and you have to defend those ui decisions, even when each of your support staff wants something different, and knows there's nothing preventing you from changing it to exactly what they want. even when exactly what they want contradicts with what the person sitting next to them wants.
Yeah, the last 1% takes 1000% of the time.
Also nice try Zendesk ;)
As someone who is currently in the process of integrating Zendesk: It seems to have a moat, however directed in the wrong direction. This integration might be my first project failure in the last decade which included two SAP integrations.
There is just so much clunk and developer hostile stuff going on that I would rather just not deal with it anymore.
It's that last S in SaaS that builds the moat around companies like this. Even if you can make the Software, it's on you to provide the Service as well if you go that route. Quite a lot of companies simply don't want to self-host yet another thing and make their IT group even larger (or more burdened). It's rarely the Software or the Platform that's the selling point as much as the Service.
or, as you have been able to do for decades, use an existing open source solution? Zendesk’s value isn’t in the CRUD operations. If a company doesn’t want to pay for Zendesk, don’t ask Claude to build a replacement… use a good free solution, that has years of domain expertise built into it.
If you have an internal CRM and are maintaining a Zendesk integration to create support tickets within your product, you've likely already done half the work needed to instead create those tickets in your own internal CRM tool yourself. This makes a lot of sense.
As someone who's had to maintain a Zendesk integration such as this for a large app it's hard to understate the benefit of having all that support info so close to the rest of your user's data. I have seen a huge amount of effort go into trying to get just the right balance of data in and out of ZD. Also helps alleviate concerns with sharing too much customer data with a third party.
This definitely isn't the right call for everyone, but there's a lot of upside if it can work for your organization.
86% AI generated text. Seriously, does nobody write anymore?
The first sentence being "I was not planning to write this article" is the cherry on top.
Great question! This is certainly a topic that touches on many important aspects of our modern world, and there are a lot of nuanced perspectives worth considering.
The problem — it's not that they don't write anymore; it's that they don't think anymore.
It sucks. The dumb diagrams as well - do we really need two boxes and an arrow to repeat what was described in a simple line of text just before?
Wow, yes, those are a pisstake. I hadn't seen them when reading the article because I used reader mode.
it's really sad. the internet is slowly dying (along with humans using their brains, it seems). On twitter, any time anything interesting is posted (which also is usually "enhanced" with AI) 80% of the responses are all very clearly just claude/gpt replies. what is the fucking point? whats the end game for these accounts?? i HATE that i have to sit there parsing through all of this cruft. Fuck this whole article could've been so much shorter, and more concise if a human just sat there, timeboxed themselves, and wrote!
No, it's just that people who were too lazy to write anymore than a paragraph can now paste that paragraph into a chatbox and get a full blogpost. The only issue is that it still contains only as much content as was in the single paragraph, the rest filler.
Actually LLMs are like the ultimate laziness enabler. But as people adjust their expectations this stuff will no longer look impressive, since people will know it's just AI slop, and the lazies will fade back into obscurity.
Hahaha totally relatable. Love Zendesk but they kept charging and adding more features I don't need.
I built https://pointanswer.com/ for myself to host 3 of my own SaaS instead of paying more than $100/mo for simple helpcenter. I'm the only one using it with no customers but it's still way better than paying for Zendesk.
But did spent too much time on it as I built in pre-Claude era.
My personal experience: For many years I used Zendesk to manage support and host my documentation. It's a powerful platform with help center structure that I liked liked the most. I was paying about US$30/mo for OnVoard.
When I use Zendesk for my second SaaS business, RenderKu, it costs me $70/mo. This amount is more than 3 Hetzner servers I'm paying to host the whole infrastructure. And I'm only choosing Zendesk mainly for hosting documentation.
At this point, when I was planning to start my 3rd app business, the thought of forking another $70/mo for hosting documentation made me rethink my options. I've eventually come to the conclusion it would be better for me to start PointAnswer and use it as helpcenter for myself since I only needed simple and affordable helpcenter.
Isn't this a perfect case for why SAAS stocks have been puking all year?
I certainly don't think Zendesk's core business is threatened here, and I have no desire to replace Zendesk with a custom solution. But Zendesk's ability to upcharge, sell adjacent products, and pass on cost increases is hugely threatened. If PE was planning on a decade of consistent cost increases and more land, that dream seems materially threatened!
I have a friend who’s doing this for himself. He owns an AC business. He has need for tools but does not like plunking down money for a feature set that moves him in a direction he doesn’t wish to go. Solution? Create a bespoke internal system of one off apps for his OWN business using an LLM.
He’s not a software developer, he has no concept of software maintenance or security.
I’ve been watching and it’s interesting to me because I would not be surprised that he’s not alone as a small family business. Many probably feel liberated from company’s that would enforce a certain cookie cutter shape.
Does this mean AI is shifting towards contractor jobs more? Does it mean a huge security issue brewing? Both? Maybe business owners turned SWVibers like him will swing back to an off the shelf option once pouring more effort into 3am-my-stuff-is-broke scenarios becomes more of a chore than it’s worth.
I feel like there are a million billion green field projects brewing that will soon turn brown for one reason or another.
> Maybe business owners turned SWVibers like him will swing back to an off the shelf option once pouring more effort into 3am-my-stuff-is-broke scenarios becomes more of a chore than it’s worth.
I mean, that's what, as an industry, we're all desperately hoping for, but, well, shit. ChatGPT-5.5 is quite capable, so if the business owner is disciplined and prompts it with thought, I'm not so sure those greenfield projects are going to turn brown. Hell, if I was an AC business owner, would I rather pay a software startup who think they know the AC business, or pay another AC business owner to use the software they wrote?
"Your job isn't going to be taken by AI, it's going to be taken by someone using AI" -Jenson Huang
Problem is, that person using AI is from outside our field. (Not a problem for the AC business owner who has a new product to sell! Good on them, if they choose to go that route.)
It would be an interesting read if they publish how things are working after a month or so. The first 90% is always is interesting and quick stuff.
On a different note, what Zendesk did is pretty much regular play now, hopefully some serious competition will keep that in check.
We ended up doing that ourselves, moved completely to an internal system that integrates well with our own administrative capabilities. It has been such a breath of fresh air, and took us about the same time to accomplish.
The final straw was attempting to move Zendesk down to one seat to have historical/archive capabilities for a period of time, and they couldn't even manage that process properly.
At some point, some people will just publish open source single-shot LLM prompts that define a clean-room spec for every over-priced product.
Why not make an open source alternative to the product?
Because everyone's needs are just a little different, and collaborating takes maintenance. Forks are free, merges are expensive.
that's actually NOT what you want to do. You want to build out a clone for these SAAS product that don't have the features you don't need, eliminating footguns and accelerating the fast path to user success.
Large companies may have difficulty embracing this strategy because software is a cost center, and not a revenue center, for them. The returns to efficiency gains are really hard to measure.
Yes, I want to build a clone for the SAAS product my company uses.
And if someone has gone to the bother to make a prompt that helps my LLM build one, and it either already doesn't have the features I don't need, or I can have my LLM start by trimming the features I don't need before it builds it...
Then I think you're saying the same thing I'm saying.
"Build an Oracle clone in Rust. Make no mistakes."
Huh, the LLM deleted itself.
I guess that is a valid solution to that prompt.
Big win for all the developers out there. This is how simple it can be, I loved every word in this
It certainly is true that many SaaS subscriptions can be replaced with custom written(supercharged by AI) tools. The key question however remains “Who are you going to call when something goes wrong?” Or “who is going to maintain this software?” Because essentially that’s what SaaS subscriptions genuinely offloaded and let you focus on your core business things.
As is most things in engineering, this is a trade-off and the line where buy vs build exists somewhere to be found and justified. Sure AI moves the line closer to build these days.
> We pulled an ultrahackathon. The team who built this did not sleep much for two days. They worked through Wednesday night, through Thursday, through Thursday night, through Friday. They ate at their desks. They wrote the spec late Wednesday evening and they wrote the cutover commit on Friday afternoon, and in between they did the work that the time between those two moments required.
Cool story but I would not want to be in their shoes. Treating your employees poorly only to justify overnight changes in business needs creates a highly toxic work environment.
Yeah that was a huge red flag. Classic power tripping CEO move to arbitrarily inject a deadline for a new project mid-sprint which then forces your team to work all-nighters and still have them deliver on all their sprint work.
If I worked there I'd be immediately looking for a new employer, I guarantee this isn't the first time they've done stuff like this over the years.
Yeah. I understand that this was an exceptional situation outside their control, but having to work multiple days straight is nothing to be proud of. It's something to apologize for and give a corresponding monetary reward for, and then see what kinds of mitigations you can take to avoid this in the future.
How much vacation and overtime pay did the team get for that effort?
Forget vacation and overtime pay, I'd expect equity at that point.
This is what I did for multiple customers now. Not Zendesk, but other expensive SaaS.
I only built the 20% they actually needed, with faster UI and better UX.
If you freelance, this is where you can make money with AI.
Thanks for this idea
>The honest answer is the boring one. We pulled an ultrahackathon. The team who built this did not sleep much for two days. They worked through Wednesday night, through Thursday, through Thursday night, through Friday. They ate at their desks. They wrote the spec late Wednesday evening and they wrote the cutover commit on Friday afternoon, and in between they did the work that the time between those two moments required.
Man, they must have gotten paid crazy overtime. Kudos!
They got paid in crazy equity that will never be liquid!
Very doubtful :)
> Man, they must have gotten paid crazy overtime. Kudos!
lol first time?
its not a US-based company, so its not unlikely
Extor-zen
Our helpdesk provider wanted a three year annual commitment at double the price we've been paying after they got bought by private equity. We are not continuing with our helpdesk provider.
The correct answer to this behavior needs to be "lol no" until these companies learn this behavior is unacceptable at any price. They could've boiled the frog at a 10% annual hike and nobody would've cared, but you double and you are fired as a vendor and that's that.
This is precisely the future of consumer software engineering (no I’m not saying anything here about building robots rockets medicine etc. at least for the next decade probably)
This idea that software developer productivity being the goal for AI companies is just not it - every piece of code you put into an LLM is a you giving these giant companies your expertise
People need to remember where the bulk of engineering money goes: consumer advertising and consumer facing applications, and don’t forget most of them have tragically bad user experiences and dark patterns because they’re trying to make the software as a service model work for investors
if I can spend a week replicating your software that has all these bullshit dark pattern features and I can replicate that for my own why why wouldn’t I?
I'm curious whether people feel like UX could at all be a differentiator here. like, sure, i can build a very quick crud app with Ai in an hour, but there are lots of UX decisions that, if not prompted in the right direction, AI just handles badly.
i guess the problem is that once a teamp goes through the process of figuring out good UX for a certain flow, which can take time, that UX then becomes trivial to copy.
Every product reveals its design if you know how to interrogate it.
So knowing a software architecture for something you use is the HARDEST part to observe and the UX is the easiest to observe.
provided you can describe what you observe, and your desired workflow that matches your need, then you can replicate it provided you understand how to test and iterate, which again is trivial to learn.
Combine:
Observations of workflows to implement and
Notional data architecture
You can create a slimmed down version of pretty much anything.
I mean this is basically every image editor compared to Photoshop.
A designer that is used to all the features Photoshop has, and then you just use the most common workflows that most people use, to be the feature bootstrap for your Photoshop alternative that’s much lighter weight and cheaper etc.
It really isn't. No-one wants to maintain these things, they'll be back on an out-sourced provider within 5 years, if they survive that long.
The devs doing these kinds projects no longer think of the code as maintainable, but disposable. It's quite the 180 for a community obsessed with reusable npm packages and not reinventing the wheel. Last year Cursor bragged about ditching their CMS in 3 days: https://x.com/leerob/status/1999513884382597485
It’s not a 180
It’s measuring time to iterate/ship which is really all that matters
Reusable was great because it used to be the fastest. The fastest now is vibe/slop > iterate > iterate > iterate > ship > iterate > ship etc…
Why maintain it?
Make a new one from scratch if needed.
> if I can spend a week replicating your software that has all these bullshit dark pattern features and I can replicate that for my own why why wouldn’t I?
Have fun Chesterton's fencing yourself into much of what other companies spent years learning and solving.
Prototypes and proofs of concept were easy even before AI.
I’m well aware
Which is why you start from the principles that were “learned and solved” in the past.
Literally none of those things that you’re talking about Chestertons fence hopping are secrets or hidden
They’re very well documented in product design, project management, cybernetics, IT etc…
There’s been thousands and thousands of articles written about all the different ways to build: functional programming, monorepo vs micro services, strong typed, memsafe etc….
there’s infinite number of framework debates that are already embedded in the LLMs that you simply need to interrogate as part of the planning process before you ship anything
It’s actually really simple.
The reality is that 99% of software engineers never gave care about any of that ever. Go ask your average software engineer about Parness information hiding or Conway law inversions.
These things are trivially easy to grasp and make engineering intuitively simple to avoid MOST problems. I am consistently teaching principal engineers, VP, senior software engineers, people who know how to write code about basic basic basic information theory. It’s honestly embarrassing.
This happened to me recently with a few providers. Their billing teams got aggressive and I said fuck it and AI coded a replacement in a few nights/weekends. For me it was: Temporal, Zapier and DocuSign
I strongly believe the build/buy equation is much different in 2026 than it was in 2024
docusign? Maybe I'm just dumb, but I would definitely want a CYA third party keeping track of all legally binding agreements.
Yeah, this is the absolute last thing I'd vibecode. This is one step above sending your LLM to represent you in front of a court when you're accused of murder.
do you really need vibecoding to clone docusign if its that easy?
My successful startups had their own internal CRM as distinguishing factor to their success, it's not hard.
One started out as an acquisition of a group on sales force, we wrote migrated to our brand new crm the first month and integrated our PBX and several other functions within the first 3 months.
It's not hard, most CRMs don't need "webscale", you can whip up a highly targeted and integrated CRM in rails in the blink of any eye.
That's why I'm pretty sure in 2001/02 I had the earliest real customer, real time, audio reviews published online in the world.
Like any decision during a startup buy/build is a situation specific decision you have to make at every point and 3rd party CRMs are fine to launch with... but, I mean, come on.