Insights from ONE Digital Tech Hub: What It Really Takes to Build and Deploy Agentic Systems

The pace of progress in AI continues to accelerate, offering businesses remarkable potential for efficiency gains — but also new challenges. At a recent event hosted by Opportunity North East’s Digital Tech Hub, Fathom Group Managing Director Richard McGowan attended a panel discussion exploring “What It Really Takes to Build and Deploy Agentic Systems.”

The conversation highlighted both the promise and pitfalls of AI deployment. Here are the key takeaways:

1️⃣ Effectiveness vs. Reliability
AI tools can often be deployed rapidly with 90–95% effectiveness — but closing the final 5–10% gap is notoriously difficult. This makes AI a powerful efficiency tool, though not yet a substitute for critical decision-making. In creative applications, 90% accuracy may be acceptable; in safety-critical industries, it’s not.

2️⃣ Data Quality Matters – “Garbage In, Garbage Out.”
According to Gartner, 85% of AI projects fail due to poor or irrelevant data. While AI can help collate disparate data sources, careful curation remains essential to ensure that outputs are meaningful and reliable.

3️⃣ Context and Interpretation
AI does not understand language or nuance. Meaning depends heavily on context, and without it, even accurate data can lead to misleading conclusions or poor decisions.

🔑 The Three Essentials for Effective AI Implementation:
🥇 Subject-matter expertise
🥈 A clear, outcome-focused plan
🥉 Reliable data

Expertise ranks first — because without it, achieving the other two is nearly impossible.

Fathom Group continues to explore the practical applications of AI across the energy and technology sectors, focusing on real-world reliability, safety, and performance.