07-07-2025
Generative AI Is Booming, But Execution Gaps Remain
Gregorio Patiño Zabala, Co-Founder & Head of Business Development - FSI unit at Pragma.
The explosive rise of generative AI (GenAI) is fueled by breakthroughs in cloud computing, smarter algorithms and scalable data storage—making the technology both innovative and accessible for businesses.
Yet many organizations overlook key technical hurdles that limit their ability to fully harness GenAI's potential. Here are five proven ways to avoid common mistakes and maximize the value of this transformative trend.
1. Identify And Prioritize A Use Case
The first step in any GenAI strategy is to identify a well-defined, high-impact business problem that the technology can address. Start by studying relevant pain points that could be solved through automation or advanced data processing.
From experience, I recommend holding discovery sessions to align stakeholder needs with feasible technology solutions. Once potential use cases are identified, prioritize them using two main criteria: business value and implementation feasibility.
Remember: Not every challenge needs GenAI. In some cases, traditional machine learning or other analytics techniques may be more effective and cost-efficient.
2. Prepare Your Data And Assess Project Feasibility
GenAI is only as strong as the data it's built on. Without a solid, evolving knowledge base, any initiative risks producing vague or inaccurate outcomes. Building a dynamic data infrastructure that grows with your business is essential.
One example: In a project we led for the insurance division of a financial group in the Dominican Republic, we helped create a conversational AI agent backed by a secure, up-to-date knowledge base. Given the constantly changing nature of insurance offerings, this empowered advisors to deliver real-time, accurate responses.
How? By querying data, extracting customer insights, generating personalized documents and ensuring strong data governance. Today, the solution supports more than a dozen core functions for the business.
3. Embrace An Iterative Approach
Implementing GenAI isn't a one-time deployment—it's a continuous process. Start with a focused pilot, adjust based on real-world feedback and scale gradually. This approach ensures reduced risk while staying aligned with business needs.
In the earlier-mentioned example, what began as a proof of concept evolved into a full solution with eight core features—from real-time data queries to pricing assistants. Each function was added incrementally based on value and feasibility.
The takeaway? Start small, iterate fast and scale wisely.
4. Build A Multidisciplinary Team For Scale And Security
To build a scalable, secure GenAI solution, companies must assemble a multidisciplinary team with each function contributing to success:
• Backend and frontend developers who ensure technical integrity and user experience
• QA engineers who test for reliability and performance
• Solution architects who align the system with business goals
• Data analysts who refine information and boost model accuracy
To reduce risk, begin by following security standards like the Open Web Application Security Project (OWASP) Top 10 for large language model applications. While many tools exist to streamline implementation, success comes from combining technical skill, domain knowledge and governance.
5. Adopt New GenAI Advances With Intention
GenAI is still an emerging technology and continues to evolve quickly. Vendors regularly announce breakthroughs, making it hard to keep up. That's why it's essential to research, evaluate and test each advancement carefully.
Organizations can adopt generative AI advances more effectively if they follow these steps:
• Establish a dedicated interdisciplinary team to evaluate new tools.
• Assess how each advancement aligns with business goals.
• Integrate human insight throughout the development process.
The goal is not to chase novelty but to adopt what creates real value—at the right time.
Bonus: Preparing For The Next Frontier—MCP And A2A
The next leap in GenAI lies in agent-to-agent (A2A) communication. Instead of standalone tools, AI systems are beginning to talk to each other, coordinate tasks and make shared decisions.
Imagine this: A virtual assistant helping with a customer return automatically checks inventory and schedules a pickup—all handled through intelligent collaboration between two AI agents.
Making this possible are new standards like the Model Context Protocol (MCP), which allow data systems and AI tools to operate together seamlessly. The real opportunity lies in building secure, flexible solutions ready to grow and evolve with your business.
When your systems can talk to each other intelligently, that's when AI becomes a true partner—not just another tool.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
