Companies Manage AI Coding Agent Budget Costs

As AI coding agents become more integrated into enterprise workflows, companies are facing the challenge of managing the associated costs. Technology leaders from Replit, Kilo Code, and Symbotic discussed their approaches to controlling budgets while leveraging the efficiency gains from these AI tools at the VB Transform 2026 conference.

At Kilo Code, co-founder Emilie Schario noted that engineers now spend only about 1% of their time directly writing or reading code, with AI agents handling the rest. This shift raises questions about which tasks are safe to delegate, error correction responsibilities, and managing multi-model architectures. The significant increase in token expenses has led to scrutiny over whether it represents genuine progress or simply inflated IT spending.

Symbotic, a warehouse automation company, focuses on directing AI's capabilities, according to distinguished engineer Jared Go. He emphasized that while AI excels at greenfield development (creating new codebases), brownfield tasks (maintaining and updating existing code) still require human oversight for critical product decisions. Replit employs a more cautious strategy, where each code change request is assigned a risk score. Low-risk changes can be auto-merged, while higher-risk ones undergo human review, a process head of product engineering Amol Jain described as "human on the loop, not human in the loop."

Supporting multiple AI models is also a key consideration. Kilo Code offers access to over 500 models, allowing businesses to select more affordable options for certain tasks. For instance, expensive frontier models might be used for initial architecture planning, followed by less costly open-weight models for subsequent development. Replit aims to optimize cost versus capability, making model choices on behalf of users to minimize expense and maximize output.

Cost control remains a priority. Kilo Code advises clients to use cheaper models for routine tasks, while Symbotic has implemented monthly cost tiers for employees and monitors usage trends to manage expenses effectively. The focus is on ensuring that AI investments translate into demonstrable value, moving beyond simply the amount spent to the return generated by that spend.