How IT pros can weigh their options as AI makes building software in-house easier than ever

It was the best of times, it was the worst of times—and for the software industry, which has faced continual discussions about its survival in the era of AI, it was well, just the times.

Anxiety about the fate of the traditional software industry has reawakened after a less-than-stellar week on the markets for publicly traded software companies, with investors questioning if AI can really disrupt a once-stable market.

Rough week. On Jan. 30, Anthropic unveiled new plugins for Claude Cowork, its desktop agent tool, which would help automate sales, legal, finance, and marketing functions.

The announcement sent Wall Street into a spiral. Shares of Thomson Reuters, ServiceNow, and Salesforce, among others, tumbled the following week. Additionally, the Wall Street Journal reported, two S&P indexes that track the performance of software and tech stocks lost almost $300 billion in market value that same week.

Déjà vu? The current industry chatter may sound familiar. Last year, the emergence of agents had some folks, including Microsoft chairman and CEO Satya Nadella, questioning whether the software-as-a-service (SaaS) model as we know it would be able to coexist with AI.

Daniel Bardenstein, CEO and co-founder of cybersecurity startup Manifest, told IT Brew that fresh concerns about the software industry are focused on the possibility that one AI provider could upend the need for different SaaS tools. Bardenstein framed the industry concerns around software vendors as “overhyped doomsday predictions,” comparing the hype of AI tools to that around self-driving cars, a common parallel drawn within the industry.

“There’s a lot of discussion around what they can do and what the future could look like, but then those predictions get pulled back a little bit when the first incident starts to happen…and so we’re in the same place with agents [as self-driving cars], granted that risk is a little bit lower,” Bardenstein said.

“We’ll see companies invest in building some things in-house or maybe outsourcing some stuff through Anthropic, but to the extent that it’s going to displace large swaths of the SaaS market that have been building purpose-built tools for five or 10 years, I don’t see that happening anytime soon,” he said.

To build or to buy? For IT pros mesmerized by the idea of building in-house software with the help of AI tools like Claude Cowork, such tools may confer benefits.

“You can avoid the vendor lock-in, plus now you can build highly-customized applications,” said Qilong Wang, CTO of web app security company NightVision.

However, Wang said building in-house may not be the optimal path for every company. Companies that don’t have proper guardrails in place to ensure the security of internally built tools, or a proper monitoring system for them, may be better off relying on traditional software incumbents.

“If you don’t have all those proper control mechanisms in place for monitoring, for operational resiliency, for functional testing, for security, then it would still be a little bit risky to use vibe-coding to build your in-house system, especially those in-house systems will be needed to support business critical operations,” Wang said.

Should software companies be shaking in their boots? For software vendors, remaining relevant in the upcoming times is going to be hard, according to Rimini Street Global CIO Joe Locandro. He recommends they revamp their strategy for the AI era by re-evaluating how they charge enterprises for their services or seeing how they could satisfy consumers’ desire for fast innovation.

“There’s an opportunity cost,” Locandro said. “I either spend [money] on my existing vendors and hope that they’re driving the innovation, the speed, the flexibility…[or have] a platform where I can mix and match and do it whenever I want, customized to whenever I want, and I don’t have to follow that.”