Using Agents in Production: Past Present and Future // Euro Beinat

Why It Matters

Understanding the low production conversion rate reveals critical gaps in MLOps and agent reliability, guiding firms on investment priorities for scalable AI automation.

Key Takeaways

• Prosus deployed 7,949 AI agents across operations
• Only 15% agents reached production readiness
• Majority serve as experimental learning platforms
• Conference highlights lessons for scaling AI agents
• Future focus on generative AI-driven agent autonomy

Pulse Analysis

The surge in AI‑driven autonomous agents reflects a broader shift toward hyper‑automation in enterprise environments. Prosus’ deployment of nearly eight thousand agents, disclosed at the Coding Agents conference, provides a rare quantitative glimpse into how organizations are experimenting with AI at scale. While the headline figure sounds impressive, the fact that only 15% of those agents have transitioned to stable production underscores the steep technical and governance hurdles that still dominate the field.

Key challenges highlighted during the virtual summit revolve around MLOps maturity, data quality, and continuous monitoring. The majority of Prosus’ agents remain in a learning phase, serving as testbeds for model iteration, feedback loops, and risk assessment. This experimental approach, though costly, yields valuable insights into failure modes, latency constraints, and integration complexities with legacy systems. Industry participants emphasized the need for robust version control, automated validation pipelines, and clear hand‑off protocols between data scientists and operations teams to improve the production conversion rate.

Looking ahead, the conversation turned to generative AI and its potential to endow agents with greater autonomy and creative problem‑solving abilities. Experts predict that next‑generation agents will combine large‑language models with domain‑specific knowledge graphs, enabling more nuanced decision‑making without extensive human supervision. As enterprises seek to monetize these capabilities, the focus will shift from isolated pilots to orchestrated agent ecosystems that can adapt in real time, driving new business models and competitive advantage.

Original Description

March 3rd, Computer History Museum CODING AGENTS CONFERENCE, come join us while there are still tickets left.

https://luma.com/codingagents

Agents in Production Virtual Conference.

Abstract //

Prosus has shipped over 7949 agents. 15% have worked. The rest have been learning experiences. Let's talk about what we have learned, and where we see things going.

Bio //

I am a technology executive and entrepreneur in data science, machine learning and AI. I work with global corporations and start-ups to develop products and businesses based on data science and machine learning. I am particularly interested in Generative AI and AI as a tool for invention.

A Prosus | MLOps Community Production