All enterprises are in the software business, but few of them want to be.
Your executives do not care about your clean architecture. To the board, your entire engineering team is just a massive spreadsheet liability they are actively trying to shut down and outsource to a vendor. But AI just gave us the ultimate leverage to fight back.
Welcome to Lead Prompt // executing leadership from the root. I’m your host, John Collins.
I work at a large enterprise that is constantly trying to shut down my projects and my teams. The reason they are doing this is because they see software as an overhead, given that it is not their core business. Like many large enterprises, software acts as an enabler of the core business, but it is not revenue-generating in and of itself.
They are not making money from selling software licenses, because they are not a software vendor. Instead, they are running applications, and even building applications, to support their non-software revenue streams. For them, software is a necessary evil, a line item on a spreadsheet that they constantly wish they could shrink or delete entirely.
During my long career, I have had the experience of working for software vendors where their applications were their core business, and for other companies where it is not. In this second category, I can promise you that software leaders have way less influence on the direction of the company, so choose your career path wisely. If you are not building the product that brings in the money, you are viewed as a cost centre.
I stick it out at an enterprise that resents software because I enjoy the technical problems that my team and I are solving together. I like the team itself that I have hired and mentored over several years now. Put simply, I am loyal to my team more than the enterprise itself. But protecting that team means understanding the financial anxieties of the executives above me.
Build versus buy
In such enterprises, their frustration towards having to build and maintain their own applications will often drive their leaders to buy solutions instead from external software vendors. The problem with buying instead of building, of course, is that you exchange the problem of having to manage a software engineering team with the problem of managing a vendor, and vendor management is a whole specialist field in itself.
When you hand the keys to a vendor, that vendor has a lot of leverage over you. They can increase their rates for customizations, which are always required at large enterprises, or they can increase their license fees. You will also find that you will have some influence over their product roadmap as a paying customer, but you do not fully control it. In the worst-case scenario, a vendor can hold your business data hostage, making switching costs extraordinarily high, which inevitably results in vendor lock-in.
Regardless of all of this, I still see large enterprises happily paying tens of millions of dollars to software vendors because they simply do not want the hassle of building and maintaining their own software solutions. To them, the massive vendor contract feels like a predictable business expense, while an internal engineering team feels like an unpredictable management headache.
The Calculus Has Changed: Enter AI
For decades, the "buy" argument won because software engineering was seen as a linear cost. If you wanted to build more features or maintain a larger internal system, you needed to hire more engineers. Headcount ballooned, and so did the enterprise's resentment.
But we are living through a fundamental shift that is completely rewriting the build-versus-buy playbook. Artificial intelligence, specifically generative AI and agentic coding workflows, is disrupting the unit economics of software engineering. AI offers a highly practical solution to the enterprise's biggest objection: the high operational cost of in-house development.
Enterprises are now realizing that they can maintain bespoke, highly customized internal software without the bloated headcount they used to fear. We are seeing massive organizations report incredible efficiency gains by integrating AI coding assistants; for instance, Walmart reported saving over 4 million developer hours annually through AI-powered development automation. That is the equivalent of thousands of full-time developers shifted away from boilerplate work and refocused on high-value business logic.
When a company arms its existing engineers with AI, it isn't just writing code faster. It is scaling context. Instead of the old formula where more output required more developers, the new reality is that an AI-augmented engineer can manage vastly broader domains. The AI handles the syntax, the boilerplate, the test generation, and the routine debugging, while the human engineer focuses on complex architectural decisions and aligning the software with the company's core business needs.
Reclaiming the Tech Stack
Because of this AI-driven efficiency, the threshold for building bespoke solutions has been drastically lowered. Why would a company pay tens of millions in licensing fees for a bloated, off-the-shelf platform that barely fits their processes, when a small, AI-empowered internal team can build a perfectly tailored solution for a fraction of the cost?
We are already seeing the leading edge of this reversal. Companies are starting to rip out monolithic vendor contracts and replace them with internal tools built with the help of generative AI. For example, Klarna famously replaced its massive Salesforce CRM deployment with a custom-built internal platform, achieving with a small team of AI-augmented engineers what would have previously taken a massive roster of traditional developers.
This is the ultimate counter-argument for the enterprise that resents software. You no longer need an army of hundreds of engineers to maintain your internal systems. You need a highly skilled, specialized strike team that uses AI as an engineering multiplier.
Securing the AI Workflow
However, as an engineering leader, you cannot just tell your developers to start pasting proprietary enterprise code into public AI chatbots. If enterprises resent software costs, they resent data breaches and intellectual property leaks even more.
The rapid rise of AI coding tools introduces the massive risk of "shadow AI", namely developers bypassing corporate security to use unsanctioned AI tools. To make in-house AI-assisted engineering viable and safe, leaders must adopt a platform engineering mindset. You have to build safeguarded sandboxes where your team can leverage large language models without exposing corporate secrets to the public internet.
By standardizing these tools and integrating secure AI assistants directly into the developer environment and fine-tuning local open-weight models on the company's own codebase, you maintain oversight and developer accountability. AI becomes a governed, secure part of the CI/CD pipeline, automating the generation of test scripts, analyzing infrastructure bottlenecks, and enforcing coding standards.
The Takeaway
So, how do we use this to our advantage? If you are a software leader working in an enterprise that views your department as an expensive overhead, AI is your ultimate leverage.
Stop trying to convince the CFO to love software for the sake of software. Instead, speak their language: cost containment and risk mitigation. Position your team not as a growing expense, but as a lean, AI-augmented unit that can eliminate the company’s dependency on predatory software vendors. Show them that by investing in secure AI developer tools, you can build exactly what the business needs, retain full ownership of the data, and do it for less than the cost of renewing that multi-million-dollar vendor license.
Software doesn't have to be a necessary evil. With the right AI strategy, it can be the most cost-effective advantage the enterprise has.
I'm John Collins, see you next time on Lead Prompt.
References
- Walmart doubles down on AI with broader rollout of coding tools - https://www.ciodive.com/news/Walmart-generative-AI-agents-coding-tool/740545/
- Klarna-Salesforce-Workday Partnership Called Off Amidst Major Gen-AI Overhaul - https://www.salesforceben.com/klarna-salesforce-workday-partnership-called-off-amidst-major-gen-ai-overhaul/
Download audio
File details: 11.0 MB MP3, 8 mins 21 secs duration.
Title music is "Apparent Solution" by Brendon Moeller, licensed via www.epidemicsound.com
