AI coding agents are blowing through budgets — Replit, Kilo Code, and Symbotic explain how they're managing it
At Kilo Code, engineers are reading or writing code themselves only about 1% of the time now, according to co-founder Emilie Schario — the rest is agents. That shift is forcing new questions onto dev teams: which systems are safe to hand over, who cleans up when models goof up, how to support multi-model architectures, and whether skyrocketing token bills mean real progress or just burned IT…
Kilo Code, Replit, and Symbotic are grappling with the costs of using AI coding agents, but are finding ways to manage them. Kilo Code co-founder Emilie Schario says that engineers now use the agents for nearly all coding tasks, leaving engineers to focus on reviewing and refining the output. Kilo Code supports over 500 models and recommends using expensive models for initial design work and then switching to less expensive open-weight models for the rest of the project.
Replit takes a more conservative approach, having humans review pull requests and assign risk scores, with low-risk PRs being merged by the author and others going to human reviewers. Symbotic has set up per-month cost tiers for employees and created a tool to monitor PRs and usage trends, allowing them to move users up or down tiers as needed.
All three companies emphasize the importance of using multi-model architectures and carefully considering model limitations and cost versus capability. Ultimately, they believe the key to managing AI coding costs is focusing on value and ROI rather than simply tracking spending.
Written by urgent.news from VentureBeat's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
