Making AI Work: A Buyer’s Guide for Leaders — Part 2

‍

About The Episode:

Welcome to the 100th episode of Generation AI, hosted by Ardis Kadiu and Dr. JC Bonilla. In this landmark episode, Ardis and JC dive into the second part of their three-part AI Buyer’s Guide series, helping higher ed leaders and enrollment teams make smarter, more strategic decisions about AI investments. This episode shifts from theory to tactics, offering a step-by-step framework for evaluating AI vendors, mapping tools to real-world use cases, and avoiding the common pitfalls that tank 95% of AI pilot projects. If you're wondering how to buy AI that actually delivers ROI — this is the episode for you.

‍

Key Takeaways

• Don’t chase features — solve real problems: Effective AI buying starts with a clear use case, not a shiny demo.
• Map your AI options into three clear categories: Assistants, copilots, and workflow agents — each with different levels of integration and ROI.
• Avoid the 95% failure rate trap: Most AI pilots in higher ed fail due to poor problem definition, lack of checkpoints, and internal readiness issues.
• Use a 5-question framework to evaluate vendors: Focus on use case fit, data gravity, action surface, adoption, and time-to-value.
• Build vs. Buy? Choose wisely: Unless you’re ASU, most schools are better off partnering with specialized AI vendors than building from scratch.

‍

Episode Summary: The AI Buyer's Playbook – Part 2

What’s the framework for buying AI in higher ed?

Ardis and JC introduce a pragmatic framework to help institutions cut through the hype and evaluate AI tools through a strategic lens. They recommend grouping AI products into three tiers:

1. General Assistants (e.g., ChatGPT, Gemini): Easy to deploy, fast time to value, low integration needs.
2. Copilots (e.g., Microsoft 365 Copilot, Canvas AI): Embedded tools that enhance workflows within existing platforms.
3. Workflow Agents (e.g., Element451 Bolt Agents): High-autonomy tools that perform end-to-end actions across systems like CRMs and SISs.

Understanding where a tool falls on this spectrum helps buyers align expectations with investment levels, required integrations, and expected ROI.

‍

Why do most AI pilots in higher ed fail?

AI tools are often brought in with excitement but no strategy. Ardis and JC identify four common failure points:

• Starting with the tech, not the problem
  Too many campuses jump into AI based on features, not on strategic problems like enrollment yield or advising load.
• Shiny demo syndrome
  Demos are built to impress — not necessarily to solve your problem. Build your own demo script based on real use cases.
• Lack of checkpoints
  Without defined success metrics and short-term adoption goals, many projects stall or get shelved prematurely.
• Organizational readiness
  If your staff, data, or governance structures aren’t ready, even the best tools will underdeliver.

‍

What are the five must-ask questions for any AI vendor?

To make smarter AI purchases, Ardis and JC propose a five-question rubric you can apply to any vendor:

1. What’s the use case fit?
   Ask: “Show me how your AI solves this specific problem — no slides, just outcomes.”
2. Where does the data live (data gravity)?
   Can it read and write directly to your CRM, SIS, or LMS? Or is it stuck behind CSV imports?
3. How much action can it take (action surface)?
   Is the AI only suggesting next steps, or can it autonomously execute workflows?
4. How easy is adoption?
   What training, documentation, and onboarding support will you get? Does it integrate with your current tools?
5. What’s the time to ROI?
   Ask for real metrics: how long until others saw value, and what adoption rates look like in similar institutions?

These questions form the foundation for a vendor scorecard, enabling apples-to-apples comparisons that reduce bias and increase buying confidence.

‍

Should you build AI tools in-house or buy from a vendor?

While building internally might appeal to tech-savvy institutions (like ASU), Ardis warns most schools lack the sustained budget, product discipline, and maintenance capabilities to pull it off long-term. A few proof-of-concept projects might be fine, but without clear ownership and long-term support, in-house builds tend to fizzle when key staff leave or priorities shift.

Instead, the data shows that partnering with specialized vendors leads to higher success rates — over 67% according to the cited MIT study, versus just 20–30% for internally-built solutions.

‍

‍

Connect With Our Co-Hosts:
Ardis Kadiu

https://twitter.com/ardis

‍Dr. JC Bonilla

https://twitter.com/jbonillx

‍

About The Enrollify Podcast Network:
Generation AI is a part of the Enrollify Podcast Network. If you like this podcast, chances are you’ll like other Enrollify shows too!  Some of our favorites include The EduData Podcast.

Enrollify is produced by Element451 — the next-generation AI student engagement platform helping institutions create meaningful and personalized interactions with students. Learn more at element451.com.