![](\"https:\/\/fortune.com\/img-assets\/wp-content\/uploads\/2026\/09\/paul-fipps.sm_.avif?w=2048\")
<\/p>\n<\/p>\n
There\u2019s an idea gaining traction in enterprise technology: companies can save millions by using AI to build their own software. Forget buying a platform like SAP, Workday, or HubSpot. Instead, CIOs and their teams are creating custom apps with Claude Code and other tools.<\/p>\n
I get the appeal because I spent years in the CIO seat. Earlier in my career, the chairman of a beverage company I was working for asked my team to build a custom analytics dashboard. We had the talent. We had the resources. We figured, why not?<\/p>\n
Four months later and setback after setback, we had to confront a harder question: Even if we could build it, should we? Every month we spent building it was time my team wasn\u2019t spending on other priorities that could create more value for the company. So we killed the project and bought a solution from a software company. <\/p>\n
Fast-forward to today and I\u2019m on the other side of the equation \u2014 I\u2019m now a leader at a global software company that helps huge enterprises manage workflows for HR, customer service, finance, and more. I interact everyday with customers who are debating building their own software with AI or working with us.<\/p>\n
They know I have a product to sell. They also turn to me because they trust my expertise and experience on the ground. My advice: there are times to leverage AI to build your own software, but there are also cases where leaning on outside expertise and buying software is a no-brainer.<\/p>\n
Before taking the plunge, it\u2019s worth asking a few questions to figure out whether building or buying software is right for you. Your company\u2019s prospects, not to mention your own job as CIO, could be on the line. <\/p>\n
At first glance, building instead of buying software might look like an easy call. You have all the engineering talent you need to crank out an app or platform, and the team is raring to go.<\/p>\n
But is building software a smart move, or does it steal cycles from your core competency?<\/p>\n
I see this with some of the most sophisticated engineering organizations in the world. One of our customers, a global technology company with hundreds of thousands of employees, has no shortage of people who know how to build stuff. Instead of wrestling with creating in-house software for billing and HR, however, they decided to keep their focus on innovation. <\/p>\n
For any business today, even high-tech players, it\u2019s all about leaning into your expertise, not building something totally unrelated to your business.<\/p>\n
For CIOs looking to build software, the savings pitch usually starts with the recurring subscription they\u2019ll cut. In some enterprises, that could mean millions of dollars a year. But this doesn\u2019t account for what it costs to develop, run, and maintain the replacement.<\/p>\n
For context: Depending on how complex their business is, we\u2019ve seen that companies building their own LLM-based software solution will typically spend five to 10 times more than using our workflow automation platform. <\/p>\n
Initial development costs are only the tip of the iceberg. Maintenance is the real expense, as well as the ongoing cost of keeping it secure, current, and running as the business changes. <\/p>\n
I\u2019ve seen this math play out firsthand. At a large financial services company, the CISO wanted to use an LLM to rewrite a core system. That process could easily take 18 to 24 months, all to save the equivalent of 0.5% of the company\u2019s annual operating budget.<\/p>\n
Getting software to work properly is tough enough. For companies that choose to build it, security and governance is a whole other ballgame with massive risks.<\/p>\n
Cybersecurity in the AI era has never been more complex, and the stakes have never been higher. Building something that works on day one isn\u2019t a problem. The real test is defending it on day 1,000.<\/p>\n
Even when there\u2019s no ill intent, your own AI agents can go rogue, doing exponential damage before anyone notices. Roughly half of orgs have seen AI agents exceed their permissions. Take car rental management platform PocketOS, which got its production database wiped by a coding agent in nine seconds.<\/p>\n
That\u2019s why it\u2019s critical that CIOs do an honest assessment of their team\u2019s capabilities. Is your platform secure in the face of evolving threats? Is it auditable if regulators come knocking? Are you prepared to track permissions? Can you ensure your software complies with shifting regulations and requirements? <\/p>\n
There\u2019s another question I\u2019ve started asking in every build vs. buy conversation: What happens when your CIO leaves? <\/p>\n
The average tenure for CIOs is about 4.5 years. A custom platform typically takes two or more years to build and longer to mature. The person who designed it, who made the architectural decisions, who knows where every integration lives, may very well be gone before the system is actually running well.<\/p>\n
A CIO I spoke with recently described a prior job where her predecessor had built a custom platform the entire organization depended on. When he left, so did the institutional knowledge. No documentation. No support team. No vendor to call. Just senior leaders staring at something they couldn\u2019t explain and couldn\u2019t afford to shut down.<\/p>\n
All that said, building software rather than buying makes sense in the right contexts. <\/p>\n
A CDIO I know at one of the world\u2019s biggest shipping and logistics companies took this approach. He\u2019s building in-house with AI, but being very intentional about where he puts resources. He only greenlights projects that differentiate the business and tap into its \u201csecret sauce\u201d \u2014 harnessing decades of internal data and other institutional knowledge to optimize shipping routes.<\/p>\n
And he also knows when to leverage AI by buying from a vendor. Putting AI-powered platforms to work on your existing workflows \u2014 like onboarding new employees or processing time-off requests \u2014 is a great place to start. These workflows might not be sexy, but they have clear guardrails and offer a direct path to ROI. <\/p>\n
AI has made it easier than ever to build. It hasn\u2019t eliminated the need to choose where you should. I\u2019ve sat on both sides of the table \u2014 as the CIO eager to build his own software, and now as a president at a company that sells software. The decision to build or buy isn\u2019t always an easy one, but for CIOs today that choice may be more important than ever and is worth careful consideration. <\/p>\n
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of <\/em>Fortune.<\/em><\/p>\n This story was originally featured on Fortune.com<\/p>\n #gutcheck #questions #leader #building<\/p>\n","protected":false},"excerpt":{"rendered":" There\u2019s an idea gaining tracti… <\/p>\n","protected":false},"author":1,"featured_media":5581,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[941,8662,1552,4081,8166],"class_list":["post-5580","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-finance-news","tag-building","tag-gutcheck","tag-leader","tag-questions","tag-software-as-a-service"],"yoast_head":"\n