In the rapidly evolving landscape of artificial intelligence, the concept of AI agents building and improving other AI agents is gaining significant attention. Alfonso Graziano, AI Technology Lead at Nearform, gave a compelling overview of this approach in a talk titled “Building Agents.” The central idea revolves around creating a self-improving system in which the AI agent is not only the agent of development, but also the tool to enhance itself.
Building Agents: Nearform’s AI Approach — From an AI Engineer
Visual TL;DR. Trusted AI agent challenges solved with Alfonso Graziano’s expertise. Alfonso Graziano’s expertise covers agent building agents. Build the agent using agent harness engineering techniques. Harness engineering methods integrate SME and human feedback. Harness engineering methods lead to measurable improvements. SME and human feedback enable visible improvements. Visible improvements envision the future direction of AI.
The challenge of trustworthy AI agents: Industry recognition that AI development is unreliable and chaotic
Alfonso Graziano’s expertise: AI technology lead at Nearform and author of Learning AI-Native Software Engineering
Agent-building agents: AI agents systematically create and improve other AI agents
Harness Engineering Methodology: Nearform’s approach to systematic and reliable AI agent development
Feedback from SMEs and humans: critical for guiding and validating agent improvements
Measurable improvements: leading to more robust and capable AI systems
Future AI directions: Enabling self-improving AI development cycles
Visual TL;DR
Alfonso Graziano: A pioneer in AI agent development
Alfonso Graziano, AI Technology Lead at Nearform, brings a wealth of experience to the discussion. His work focuses on building AI agents and supporting teams implementing AI-native engineering methodologies. Graziano, author of Learning AI-Native Software Engineering, is at the forefront of finding ways to make AI development more systematic and reliable.
The challenge of building trustworthy AI agents
The presentation began by highlighting the industry consensus. Everyone wants AI agents, but the reality often comes with significant challenges such as illusions, high costs, and over-reliance on hype. Graziano illustrated this with a humorous meme depicting a long queue of “AI agents” with a list of associated issues, contrasted with a shorter, more desirable queue of “automation” focused on reliability, ROI, and scalability.
