VeloDB executive explains why AI agents fail after pilots
In short
At TechSparks 2026, VeloDB’s Regional Director for APAC said AI agents that work in pilots often break when they move into production.
He pointed to the database underneath them as a key place where the problem shows up, suggesting that better infrastructure is needed for these tools to work reliably.
What the reporting says
At TechSparks 2026, VeloDB’s Regional Director for APAC explained why AI agents that work in pilots often struggle in production. The remarks pointed to problems in the database layer underneath the agents as a likely reason for the breakdown. The brief focuses on the gap between demo success and real-world use, and on the infrastructure changes needed to make these systems work reliably.
Why it matters
At TechSparks 2026, VeloDB’s Regional Director for APAC discussed why AI agents that succeed in pilots can fail in production. The remarks linked the problem to the database layer supporting the agents, implying that data infrastructure is a major factor in whether AI systems run reliably outside demos. The comments are relevant to startups and companies building or deploying AI agents.