Are AI Agents the Future of Developer Productivity in the Enterprise?

"There's a lot of hype with the AI agents and their productivity and potential outcomes. AI Agents are quite amazing, says Eric Paulsen, EMEA Field CTO at Coder.

In this episode of the Tech Transformed podcast, Shubhangi Dua, Podcast Host and Producer at EM360Tech, talks to Paulsen about the constantly advancing role of AI agents in development environments. 

Paulsen explains how AI agents can help developers by handling simpler tasks, almost like having assistants or junior developers to assist them. Not only would this boost productivity and time efficiency, but the technology will also ensure human oversight. 

The conversation further explores how AI fits into cloud development environments, especially in regulated areas like finance, where security and scalability matter most. Paulsen stresses the value of internal AI models and points out Coder's unique role in offering infrastructure-neutral solutions that meet various enterprise needs.

AI Agents Are More Than Just Code Writers

When people hear "agentic AI" or "coding agents," there's often a misconception about fully autonomous coders. However, Paulsen clarifies, "That's a far stretch from where we currently have been, which is with just AI-assisted IDE extensions such as GitHub, Copilot, Amazon Q Developer and systems of that nature." 

Coder focuses on agentic solutions that have a human developer in the loop, emphasising Paulsen. “Think of an AI agent as a junior engineer working alongside you,” Paulsen explains. 

"If anything, it’s improving the output of the human engineer by having an autonomous or artificial or AI process. In the same development environment, working on other tasks that might not necessarily be as complex," he adds. This means developers can offload simple tasks like bug fixes or dependency updates, freeing them to focus on more complex features.

How to Scale AI Agents Securely in Enterprises?

For large financial institutions that have hundreds and even thousands of software engineers, deploying AI agents at scale requires a consistent and secure approach. Cloud development environments provide the best way to deliver and package these agents for developers.

The main concern for enterprises is ensuring data security in addition to stopping AI agents from "running wild on a laptop." Paulsen stresses the need for agents to work within an "isolated compute," with "boundaries around those agents inside of that isolated compute." 

Such a secure environment provides guardrails to synchronise and boost productivity between humans and AI while preventing sensitive data breaches or "hallucinations" from the AI.

Additionally, financial institutions are now increasingly developing their own internal AI models. Paulsen mentions, "What these institutions need is an AI agent that is trained on the internal dataset and internal LLM that is built within the firm so that it can make those decisions and return the relevant output to the data scientist or software engineer." This move towards self-hosted LLMs and internal AI infrastructure is essential for adopting enterprise-grade AI.

The ultimate message is that cloud development environments should provide the framework where AI agents are running inside an enterprise’s infrastructure. “AI agents have access to the data, and they're observed and governed by a set of security standards that you have internally,” says the EMEA Field CTO at Coder.

Takeaways

AI agents can assist developers by handling simpler...