Anyone else transitioning to 100% agentic workflow where devs manage multiple parallel AI agents/machines? What are your pain points?
retained text
At my company, we recently made a massive pivot toward a 100% agentic development model. Instead of the traditional workflow where a developer works on one task/branch at a time, each engineer now manages multiple Claude Code agent sessions/machines simultaneously. The main goal is to crank out a high volume of code and push through a shit ton of Jira tickets in parallel. While output and ticket throughput are definitely up, managing multiple active Claude Code workflows at once has created a totally new set of challenges that we’re trying to navigate. 1. Data management - skills, user scope CLAUDE.md, merging system for high and rapid volume of code... 2. Human's brain context - we are all humans at the end of the day and it is getting ridiculous to try and keep up with what is going on with 50 sessions opened across 4 machines.. Would love to hear how other teams using Claude Code at scale are handling this—what’s working, what’s breaking, and what pain points are you having?
document fetched 2026-10-25 00:35 · canonical https://reddit.com/r/ClaudeCode/comments/1xeg8xs/anyone_else_transitioning_to_100_agentic_workflow