Cisco Engineering Director on Managing a Workforce of AI Agents

Sergio Freitas, a senior director of engineering at Cisco based in Lisbon, observes that the integration of frontier models like Anthropic’s Claude Opus 4.5 has transformed engineering from a craft of manual syntax into a discipline of workflow management. Engineers now act as system architects, delegating tasks to autonomous sub-agents that monitor GitHub pull requests and Jira tickets. These agents operate within strict guardrails, though they now utilize "loop engineering" to iterate on tasks until reaching a final output without constant human prompting.

This shift brings a heightened intensity to the workday. Although total hours remain stable, the volume of output has surged, requiring engineers to balance multiple threads of agentic collaboration. This new reality demands constant attention switching rather than the deep, sustained focus once required for writing code from scratch. Despite the automation, human responsibility remains absolute; every piece of code deployed to production must pass rigorous security scans and a mandatory two-engineer review process. For those who defined their professional identity through the tactile experience of manual coding, this transition represents a significant and sometimes challenging evolution in their craft.