You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 5 minutes
Key takeaways:
- The SaaS replacement wave is real but not universal. The bar to justify a subscription is much higher than it used to be.
- Data extraction is the hardest part: building a replacement is easy. Extracting proprietary data from closed-source platforms isn’t.
- The security risk is being underestimated.
As AI-coding tools emerged, technology thought leaders and venture capitalists warned of the impending death of Software-as-a-Service (SaaS).
Now, a few years on, SaaS companies face the first nail in the coffin as C-suite executives reevaluate their SaaS stacks, driven by rising geopolitical tensions and increasingly capable AI-coding assistants.
Buy-now-pay-later titan Klarna is the most prominent example, revealing in late 2024 it was shutting down several SaaS providers, including Workday and Salesforce, in favor of a more lightweight tech stack. Klarna declined to be interviewed for this article.
Others have followed in its footsteps, with a slightly more measured approach. Media publisher 6AM City estimates saving $100,000 a month from replacing its CRM with its own AI-powered version. The agentic development framework Warp is pausing SaaS purchases in favor of agent-based and just-in-time solutions. Its CEO, Zach Lloyd, estimates saving over $10,000 a year from canceled subscriptions.
Warp moved its marketing site, hosted on AI-powered website builder Framer, to its own in-house solution, giving the team more flexibility to run A/B tests powered by agents and not be limited by the confines of Framer’s APIs, says Aloke Desai, head of product engineering at Warp.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
A company hackathon also delivered Warp a new documentation solution, which brought additional functionality that wasn’t available with its existing provider, GitBook, Desai says, such as styling consistent with its brand, connecting to telemetry, and automatically generating documentation from new releases.
A lot of the replacement of internal SaaS tools is being driven by individual employees who have been empowered to experiment with AI, says Jess Lampe, global lead technologist at Launch Consulting. Developers, in particular, are finding software lifecycle tools which were built for agile development not always suitable for the new world of agentic development, he adds.
Warp’s journey started with a few experimental use cases, which then cemented its CEO’s conviction to further invest in this shift, Desai says. “One of the worries is often, ‘oh, there’s a lot of maintenance.’ For a lot of these simple apps, it’s a pretty minimal amount of maintenance, and we were able to get something that fits in our brand and is truly a better docs product with maybe like two days of work.”
A shifting SaaS value prop
More companies are heading in the direction of Warp. A survey from low-code AI development platform Retool found that 35% of its 817 customers have already replaced one SaaS tool with a custom build, and 78% expect to build more of their own tools in 2026. Retool did not respond to an interview request.
However, not every company is looking to replace SaaS with internal vibe-coded solutions.
For Berlin cybersecurity startup DmarcDkim.com, the combination of a higher pricing cliff with SaaS and growing concerns around data sovereignty, privacy, and governance in Europe pushed it to reevaluate its tech stack. In the past year, its founder, Oleksii Antypov, has replaced roughly ten SaaS solutions with self-hosted open-source alternatives. After trying Slack and Teams, Antypov opted for Rocket.chat, and for a CRM, he picked Twenty over HubSpot and Salesforce.
Felix Godbout, an infrastructure architecture consultant, is observing a big push from the C-suite to move the application suite on-premise due to data sovereignty concerns. Enterprises often realize they can eliminate SaaS in the process in favor of automation with AI or cloud functions, but it doesn’t come without risk, he says.
“People will find SaaS on GitHub, and say ‘oh, it’s open source, let’s deploy it,’ and it just outright introduces vulnerabilities within their own environment because it’s not vetted properly, it does not hold the proper security certifications, or it introduces middleware packages that are vulnerable,” he says.
More like this
To mitigate maintenance and security issues, Antypov opted for open-source offerings, ensuring they have active codebases. Warp built a shared layer of deployment and security for the apps, so custom security features rarely need to be built per app, Desai says.
“It’s important to distinguish maintenance of the infrastructure versus maintenance of the software,” Antypov says, noting it’s around one day a month for maintenance for the infrastructure, as well as the initial upfront investment to secure the servers. “We are deliberately choosing not to maintain the software itself, because that’s where the price point wouldn’t make sense anymore.”
The reality check
It’s not building a replica or even the productionization that is challenging. Often it’s the data transfer because these SaaS tools are closed-source and have internal databases, Godbout says.
Klarna’s CEO Sebastian Siemiatkowski highlighted on the 20VC podcast that the switching cost of data is one of the big challenges because proprietary data is stuck in the SaaS provider’s data model and stored according to their setup. Once switching costs are solved, it will be a real threat to SaaS, he says.
“Once you move [data] over, well, you have a lot of things that tend to break down the road, right, because you built some forms, deep inside your ERP,” Godbout says. “You built some automated processes in system XYZ that all need to be recreated, but if you forget that little simple form, well, that doesn’t work anymore, right? Or you duplicate your data on two platforms, and now for a while you’re stuck dealing with two platforms.”
Some SaaS tools have created “artificial moats” by charging extra for interacting with and extracting data. Now, as more companies move to agentic solutions, which interact with this data, they are finding friction, making those tools most at risk for reevaluation, Lampe says.
New York • September 15 & 16, 2026
Cut through the hype.
Find what works at LDX3 New York
Vibe coding does not always result in the removal of SaaS. One of Lampe’s clients has ended up onboarding more SaaS thanks to AI, as it simplified the implementation and integration of various logging tools, he says.
Just as AI-coding tools are shifting the skills required of developers, the arrival of AI is changing expectations of SaaS. It’s not dead, but the bar is much higher to justify a SaaS subscription.
“We’re always going to make that calculus, should it be like a SaaS or an internal tool? But now it’s a much more internal tool because there’s obvious product benefits for us,” Desai says.
“Whereas five years ago, or even three years ago, I wouldn’t even have thought about an agent. I would just be like, I’m gonna go Google what the right tool is or ask advisors what it makes sense to adopt as a SaaS tool and just directly adopt it.”
