AI speeds up software development, but ownership costs remain. Learn how investment firms can balance proven platforms with custom tools to build their edge.
AI makes software faster to build, but it does not eliminate the cost and responsibility of owning it.
I have spent nearly 20 years working in alternative financial technology, and I have watched software delivery change in ways that would have been difficult to imagine when I started. The tools, infrastructure, and intelligence embedded in the development process have all advanced dramatically.
What has not changed is the instinct to assume that the latest advancement has solved the entire problem.
AI-powered code generation has materially reduced the time and cost required to build software. But writing code is only one part of delivering and maintaining an enterprise application. For financial services organizations, regulatory compliance, cybersecurity, and operational resilience remain non-negotiable.
That is why build versus buy is increasingly the wrong frame. The more useful question is where to draw the boundary between them: which capabilities should a firm inherit from a proven platform, and where should it apply engineering effort to create a proprietary advantage?
When I was a child, I watched my father work on our family’s IBM PS/2, a machine with an i386 processor, two megabytes of RAM, and a 32-megabyte hard drive that we compressed to squeeze out a little more space. When the software misbehaved, as it often did, he would tell me, “Computers only do the things you tell them to do.”
That lesson in accountability and precision has stayed with me. In the age of AI, it can seem as if the rule no longer applies. It does. The instruction is now a prompt, and the machine can respond with extraordinary speed and sophistication, but a human still initiates the work and remains responsible for the outcome. AI is a force multiplier, not an autonomous actor.
Its impact on the build phase is nevertheless significant. Work that once required a team of senior engineers across multiple quarters can sometimes reach a minimum viable product in days or weeks through structured prompting and AI-assisted development. For financial services firms facing tight delivery windows and scarce engineering talent, that compression belongs in the investment case for custom software.
It just is not the whole investment case.
The common mistake is to treat lower coding costs as a comparable reduction in total cost of ownership. AI may accelerate parts of development, but firms still have to manage the full software lifecycle:
Early in my career, deployment was physical. I worked for a software company that shipped its product as a preconfigured rack-mounted server. I remember pushing those servers on a cart up Fifth Avenue to hedge funds that needed them installed before they could see a single screen.
The cloud made deployment faster, cheaper, and remote. It did not make enterprise deployment trivial. AI represents another major leap forward, but it has not removed the institutional work surrounding the application. If anything, firms that produce code faster need to be more disciplined about the controls that determine whether it is secure, resilient, and fit for purpose.
For much of the past two decades, vendors fixed the boundary between building and buying. A commercial platform delivered a defined feature set. Anything beyond it required a change request against the vendor’s roadmap or a separate custom application that the client had to maintain.
That created a genuine either-or decision: accept the platform’s limits or absorb the full cost of building around them.
Modern platforms can move that boundary. Instead of operating as closed applications, they can expose their underlying data models, workflows, and integrations as foundations clients can build on directly. A firm can inherit the platform’s security, architecture, deployment controls, and ongoing maintenance while using its own engineers to develop the logic and workflows that distinguish its investment process.
This is the principle behind VerityRMS. Firms are not limited to the workflows Verity provides out of the box. APIs and Model Context Protocol connectors allow clients’ own tools and AI systems to connect to the system of record. Investment teams can build proprietary research workflows, custom scoring logic, and firm-specific automation without recreating the application and infrastructure underneath them.
The result is not an argument against building. It is an argument for building on the right foundation. Firms can capture the speed of AI-assisted development without reassuming every pre-build and post-build responsibility that AI does not eliminate.
For CTOs and technology leaders evaluating a new application, the AI era raises three practical questions.
A build proposal should account for more than the people and time required to produce a working application. The model should include requirements definition, architecture, testing, deployment, training, monitoring, support, upgrades, and the compliance overhead specific to the institution.
This matters because a compressed build phase can make a custom project appear less expensive without materially changing many of the costs the firm will carry over the system’s life.
Engineering capacity is valuable. Firms should reserve it for capabilities that reflect their investment edge: proprietary scoring, research logic, workflows, and integrations that competitors cannot simply purchase.
If a commercial solution already addresses the undifferentiated foundation, rebuilding that foundation from scratch rarely creates an advantage. A platform that clients can extend changes the choice. The firm can buy the common infrastructure and build the distinctive layer on top.
A mature commercial platform contains more than its current features. It also carries years of requirements refinement, user feedback, security hardening, incident response, regulatory alignment, and operational tuning.
Its client base has helped expose and resolve problems that a first-generation custom system has not yet encountered. Monitoring and alerts have been tested against real production incidents. Patch and upgrade schedules are supported by a dedicated engineering organization. Training and adoption practices reflect experience across multiple deployments.
For a financial services firm, that institutional memory deserves serious weight. A new custom system may be technically sound on launch day, but it has not yet lived through a major operational incident, regulatory examination, or period of market stress. AI can accelerate development. It cannot manufacture that experience.
Platform evaluations should therefore go beyond the current feature list. Technology leaders should ask whether the platform exposes its data and workflows through documented APIs, supports modern AI connectivity, and allows proprietary development to remain additive rather than becoming a parallel system that competes with the system of record.
AI code generation has changed software economics. It has made the path to a working application faster and more accessible, and technology leaders should take that progress seriously.
But the working application is the beginning of the commitment, not the end. Requirements, architecture, testing, deployment, governance, training, monitoring, and maintenance still determine whether enterprise software succeeds.
The strongest approach is often to buy the lifecycle that a proven platform has already solved and build the capabilities that make the firm different. With an extensible foundation, the two choices reinforce each other: the platform absorbs the common cost and risk, while the firm’s engineers concentrate on its proprietary edge.
The question is no longer simply whether to build or buy. It is what your firm should build on top of what it buys.
VerityRMS gives investment teams a secure and extensible system of record for proprietary research, with APIs and AI connectivity that support firm-specific workflows and innovation.
See how Verity accelerates winning investment decisions for the world's leading asset managers.
Request a Demo