2025 IDC MarketScape: Low-Code and No-Code Technologies
There have been an incalculable number of AI vendor pitches this year, and each one promises to revolutionize business, streamline tedious tasks into oblivion, and deliver ROI that makes CFOs weep with joy. Yet here most of us are, still trying to figure out which agentic AI initiatives actually matter and which ones are just expensive science experiments.
If this sounds familiar, you're not alone.
At a recent OutSystems customer roundtable, Howard Millar, UCLA Anderson School of Management CIO, and other OutSystems customers talked honestly about separating agentic AI substance from agentic AI theater.
The agentic AI reality check every enterprise needs
UCLA is using OutSystems to position itself as an AI thought leader in business education. Both organizations agree that successful AI implementation starts with cutting through the noise.
“For the last two years, we've heard nothing but AI,” says Howard Millar, UCLA Anderson CIO. “We want the UCLA Anderson School of Management to be a thought leader for AI.” His team manages more than a dozen generative AI agents, and they're still connecting the dots between them.
Like UCLA, different teams all over the world are adopting and using generative AI to meet their needs. However, there is little control over who is using them and how. Many are deployed with little instruction or they’re MVPs, missing key elements. That's the reality most organizations won't admit. Sophisticated enterprises are still figuring out the basics of imperfect AI solutions while vendors are out there pitching the next big thing, which is most likely to be a pilot that isn’t production-ready.
A tale of three strategies
Based on their experiences and hard-won lessons, customers at our roundtable identified three approaches that are enabling them to get real results from AI initiatives.
Strategy #1. Start with infrastructure, not moonshots
Reinventing your business overnight with AI is a fever dream of the best marketing minds. Several customers have a practical, no-drama approach to AI that works. They focus on using AI to streamline existing processes across their divisions. They use platforms like OutSystems to tap into AI capabilities that complement their existing infrastructure.
This strategy works because it builds on what you already have rather than requiring you to tear everything down and start over. Your legacy systems aren't going anywhere anytime soon, so the smart move is to use AI to make them work better today.
Strategy #2. Focus on specific problems, not general capabilities
Every agentic AI initiative at UCLA ties back to the core mission of preparing students to work with AI in the real world. That kind of focus requires searching and identifying specific problems AI can solve. When other customers talked about using agentic AI, it was clear they were using it to address operational challenges that impact their business daily. They also believe that agentic AI used wisely can also help them pull together third-party apps, and are eager to explore it.
Strategy #3: Build your data foundation first
What vendors won't tell you is that AI is only as good as your data. One OutSystems customer didn’t mince words about this reality, saying that incomplete or outdated data will deliver incomplete and outdated AI.
Before you deploy a single AI agent, you need to ask yourself what data it will use and if that data is clean, current, and compliant with regulations. In highly regulated sectors like healthcare and finance, the audit and compliance piece forms the foundation everything else builds on.
The agentic AI question nobody wants to answer
Out in the marketplace, agentic AI is the big bang. One day it wasn’t here, and now it’s a whole universe. Nvidia researchers are raving over small language models, and tech conferences are full of demos showing AI agents seamlessly handling complex workflows.
But Millar raises the question every enterprise should be asking: "The general premise says that agentic will do the handoffs in support of achieving business goals automatically, but what about figuring out how to get there?"
For most organizations, the practical implementation path remains foggy. Organizations are still mastering basic generative AI, which means jumping to fully autonomous agents might be premature for most enterprises.
What agentic AI success looks like
AI success looks like measurable efficiency gains, such as streamlining processes across divisions, and clear business alignment like UCLA preparing students for an AI-powered workforce. It includes sustainable modernization, with one organization retiring legacy systems and the other building for the future, and manageable risk through proper governance, security, and compliance from day one.
What’s missing from this list are the common complaints about agentic AI:
- Endless pilots
- Proof of concepts
- Tool sprawl without governance
UCLA and other OutSystems customers have avoided those traps by maintaining a clear focus on business impact.
Your next move
The gap between AI promise and AI delivery isn't closing on its own, but you don't need to figure it out alone. The enterprises making real progress share three characteristics. First, they choose platforms that combine AI with practical development capabilities. OutSystems Agent Workbench, for example, is a unified platform for building, orchestrating, and governing AI agents without uncontrolled sprawl or integration nightmares.
Second, they start with clear business goals that go beyond "implement AI" to specify outcomes like "reduce development time by 40%" or "retire three legacy systems by the fourth quarter."
Third, they're honest about where they are on the journey. As Howard from UCLA notes, many organizations are still too early to determine whether small language models or large language models are the right approach for agentic AI, and that's okay. Building steadily beats chasing every trend.
Ready to get results? Learn how the OutSystems platform can help make that happen.
