The Case for Low-Code Everyday AI
Custom-built AI is costly and unsustainable—low-code everyday AI offers a smarter, more scalable path forward for higher education institutions.
I started my career nearly 30 years ago as a programmer, working on student information business processes at Texas A&M University. It was an exciting time to be in IT—full of innovation, problem-solving, and the thrill of building new things from scratch. Back then, I preferred a basic text editor over commercial development tools because I wanted full control over the product and preferred a deep understanding of how the code worked. But nearly three decades later, as a CIO, my perspective has shifted completely—custom development is time-consuming, expensive, and risky from a data privacy and security standpoint. Outside system-to-system integration, building IT solutions from scratch is usually a bad idea. Today’s Dispatch delves into the reasons behind my change of heart—and why low-code development emerges as the more prudent choice in the era of GenAI.
Why it matters
Custom “everyday AI” development is expensive and an ongoing support nightmare for universities. Maintaining custom everyday AI models, infrastructure, and personnel is unsustainable and diverts resources from core IT services. Instead, universities should focus on low-code and commercially supported AI solutions that integrate seamlessly with existing systems, providing scalable AI capabilities without the long-term financial, security, and maintenance burdens of a homegrown approach.
Three reasons why low-code “everyday AI” wins
Build-It solutions are always expensive and quickly become legacy tech.
The Illusion of Cheap Development—Some technologists, often early-career, underestimate the long-term cost of custom-built software. They see internal development as “free” because there is little external spending, which ignores the hidden costs of institutional IT.
Personnel costs—Salaries, benefits, and training for programmers, data scientists, and DevOps teams.
Maintenance costs—Once built, who maintains it? To remain functional, every custom-built system requires ongoing patching, security reviews, and feature updates.
Scalability limitations—Homegrown systems rarely scale well. What works for one department often doesn’t translate across an entire institution.
Custom Solutions Age Quickly—AI is advancing at an unprecedented pace. A custom “everyday AI” tool built today will be obsolete within 12 months, requiring continuous redevelopment to keep up.
Example: Universities that built custom chatbots in the last two years struggle to match the capabilities of OpenAI’s ChatGPT or Google Gemini, which require zero internal development and provide enterprise security and support.
Higher Ed’s Track Record on Custom Tech is Clear: It’s Unsustainable—Universities have spent decades trying to maintain custom-built student information systems (SIS), mainframes, and homegrown ERPs—most of which eventually became financial and technical burdens.
Example: Many institutions persisted with COBOL-based student records systems developed in the 1970s and 1980s because of the high replacement cost. However, the longer they waited, the higher the price became. If universities struggle to maintain custom-built ERPs, how can they maintain custom-built everyday AI solutions over the long term?
The takeaway: Low-code everyday AI solutions don’t require massive internal development efforts. They allow universities to adopt the latest everyday AI capabilities quickly without getting trapped in an expensive, unsustainable build cycle.
Custom AI solutions are usually less secure than commercial solutions.
Speed over Security—When IT teams rush to build internal everyday AI solutions, they prioritize fast deployment over security best practices. The result? Unsecured, untested code running on university systems connected to the public Internet.
Key Security Risks of Custom AI Development:
Lack of Code Review & Change Management—Universities rarely have dedicated security engineers reviewing every piece of custom code. Custom-built everyday AI often lacks rigorous penetration testing and oversight.
Data Privacy Risks—AI models require vast amounts of data, and improperly managed everyday AI systems could expose student records, research data, or administrative records to security breaches.
Unknown Legal Risks—AI governance is evolving rapidly, and universities developing their own everyday AI risk unintended regulatory violations, especially under FERPA, HIPAA, GDPR, and AI Act compliance.
Universities Are Prime Targets for Cyber Attacks—Higher ed is already a top target for ransomware and cyberattacks. Any internally developed AI solution that lacks commercial-grade security increases institutional risk.
Why Commercial AI Is the Safer Choice:
Enterprise security by default—AI platforms from Microsoft, Google, and AWS have built-in encryption, identity management, and AI ethics safeguards.
Regular security updates—These platforms continuously patch security vulnerabilities, while custom-built solutions often rely on underfunded IT teams for maintenance.
Auditability & compliance—Commercial everyday AI providers already adhere to strict regulatory standards, removing the legal burden from universities.
The takeaway: Custom everyday AI development exposes universities to unnecessary security risks. Low-code everyday AI platforms offer enterprise-grade security, compliance, and ongoing protection that internal IT teams cannot match.
Higher Ed IT Left the “Build-It” Era 20 Years Ago.
The Shift from Custom Systems to Vendor Partnerships—Higher education IT has evolved beyond the days of building everything from scratch. Most institutions now rely on third-party ERP, SIS, and CRM platforms—not homegrown technology.
Everyday AI Should Follow the Same Path:
Higher Ed’s IT Strategy Today = Application Integration, Not Custom Development—Universities no longer build custom ERP systems; they use Banner, PeopleSoft, Workday, and Salesforce to manage student and administrative data.
The Role of IT Has Shifted—Instead of building new software, IT teams now focus on data management, security, and application integration.
Everyday AI Needs to Fit Into This Model—Institutions should integrate everyday AI into existing platforms rather than building new, isolated AI applications that require independent maintenance.
Avoiding “Innovation Theater”—Many AI projects in higher education exist for publicity rather than actual need.
Example: The Los Angeles Unified School District builds an everyday AI-powered student and parent chatbot with a startup. Despite numerous conference appearances, the solution failed, and the startup filed for bankruptcy.
Real Innovation Happens at Scale—Institutions should invest in vendor partnerships that provide scalable, widely adopted AI solutions instead of isolated AI experiments.
Where Custom AI Development Does Make Sense:
Faculty Research: Custom models and experimentation benefit AI innovation in research fields like biomedical engineering, climate science, and computational linguistics.
The takeaway: Everyday AI should be treated like ERP and CRM systems—universities don’t need to build them from scratch. Instead, low-code everyday AI should be used for administrative functions and student services, while custom AI development should be reserved for research and highly specialized niche areas.
What’s Next?
The message for university CIOs and IT leaders is clear: Stop trying to build everyday AI from scratch. Higher education should focus on everyday AI adoption, not everyday AI engineering.
The Smarter AI Adoption Strategy:
Adopt low-code AI for everyday automation—Use AI-powered document processing, chatbots, and predictive analytics without building custom models.
Leverage vendor partnerships for game-changing AI—Universities should maximize AI investments in Microsoft, Google, AWS, and other trusted partners.
Reserve custom AI development for faculty research—The only places where “build-it” makes sense are research and innovation, not administrative operations.
The bottom line:
Everyday AI isn’t a software development project—it’s a tool for solving problems. Universities should embrace low-code everyday AI for efficiency and only build AI where it truly adds value.