AI’s Impact on Custom Off-the-Shelf Software
AI’s Impact on COTS Software
Commercial off-the-shelf software (COTS), at its most elemental level, is defined by NIST as a software and/or hardware product that is commercially ready-made and available for sale, lease, or license to the general public. From a software stack perspective, this can go as low as the operating system (OS) level (i.e., Windows, IOS, etc.). It also includes everyday office tools (e.g., Word and Excel), integrated development environment (IDE) tools (e.g., PyCharm or Visual Studio), and business applications (e.g., Workday or SAP).
The basic selling point for COTS is “do you really need to re-invent the wheel”? Why write your own OS, word processing software, or IDE when proven solutions are available? All you need to do is click and install them and you’re up and running. It makes zero business sense for an organization to develop its own solution for managing memory on a PC or writing a document.
However, business applications such as enterprise resource and planning software (ERP) are different. You can’t just click and install these. Business software is attempting to automate and enhance productivity around an organization’s systems and processes, and therefore need to be configured and/or customized to meet their requirements. Sometimes these requirements are standard, but often they are unique to an organization.
The COTS Value Chain
About 20 years ago at Oracle Open World in San Francisco, the consistent message to partners was “move up the stack.” Oracle had entered the enterprise applications market by acquiring Siebel Systems and Peoplesoft, and they wanted their partners to stop building applications from the ground up and instead add value on top of existing enterprise applications. However, with the rise of cloud computing and SaaS, large software companies have continued to build features that extend further up the value chain. While this reduces the need for customization, it also forces organizations to adapt to platform-enforced processes, even when they prefer to use their own unique processes. This issue is particularly acute in the public sector, where a patchwork of federal, state, and local legislation dictates relatively small markets. This has created the perfect storm of high-cost enterprise business application license fees and lengthy, expensive implementations.
I have worked for companies that sell and implement COTS business applications for over 20 years as a solutions consultant, product manager, and implementation manager. I’ve written some creative responses to RFPs that required me to define whether a feature was “included/out of the box,” “configured,” “customized,” or “roadmap (x months).” Even if the 80/20 rule is generally at play here— in that 80% of the requirements will be covered by the COTS product—that 20% portion of unique features can drive up implementation costs. Depending on the size of the organization and the complexity of its business implementation, costs can range from 2x-10x of the software costs. Furthermore, these customizations will haunt an organization in future years when it is time to upgrade. A prime example was an ERP upgrade at a university back in 2013, which was estimated at $83M.
The rise of software as a service (SaaS) over the past 10 years has mitigated some of these costs because these solutions tend to be less customizable. However, this inflexibility can at best lead to decreased user satisfaction and at worst the cancellation of a project.
COTS Needs to Retreat Down the Stack
Customers should not build their own business applications from the ground up. Similar to word processing software, customers should not build basic standard features like role-based application security or a general ledger module. However, at the same time, should a large software vendor trying to cover a wide variety of industries have so much control over the look and feel and user experience of a business application? What works in one state agency might not work well for a university or a county office.
The advances in generative AI call into question the current COTS model for business applications. Consider the findings of a recent study focused on GitHub Copilot’s impact on developer productivity:
- 55% increase in developer productivity
- 60-75% increase in developer job satisfaction
- 73% felt more focused on task
When evaluating the buy vs. build decision, how might these significant boosts in productivity, focus, and job satisfaction from AI influence how customers purchase and implement COTS solutions?
What about other features in AI that could reduce costs and increase the value of an enterprise application for a customized approach?
- Legislative requirements: Many software applications require annual updates due to legislative changes. Generative AI can monitor these changes and swiftly translate them into code updates, allowing organizations to keep their software current.
- Priority Features: Vendor roadmaps often promise features that take years to materialize, if ever. Customized applications empower organizations to control their unique roadmaps and prioritize features that matter most to them.
- AI Managed Code: Customized software faces three major risks: security vulnerabilities, unmanageable “spaghetti code” from quick fixes, and accumulating technical debt. AI-managed code significantly mitigates these risks by ensuring well-organized, secure, and maintainable software. Additionally, AI will make porting software to modern platforms a straightforward exercise.
Given the trajectory of AI and its influence on software development, there appears to be a sweet spot yet to be served by the market between low code application platforms (e.g., Appian or Power Apps) and more full-fledged ERP/CRM solutions/platforms such as Workday, Dynamics, and Salesforce. This might mean focusing on the foundational aspects of business software: compute, storage, security, workflow, value-added data models such as GL, or functionality such as fraud detection. With this evolution, customers could ideally realize decreased licensing costs for enterprise software applications and decreased implementation/upgrade costs in the future.
-Kevin Meldorf, VP of Strategy
