Dev Interrupted Retrofit or reimagine? Developer environments for humans and agents | Ona’s Matt Boyle
40 snips
Mar 31, 2026 
Matt Boyle, Head of Product, Design, and Engineering at Ona — former Gitpod leader focused on cloud dev environments and agent security. He talks about reimagining ephemeral workspaces for both humans and AI, kernel-level runtime controls like Project Veto, integrating agents with internal tools, and the decline of the traditional desktop IDE as web workspaces and agent loops reshape developer workflows.
AI Snips
Chapters
Transcript
Episode notes
Human Workspaces Are Agent Accelerants
- Cloud dev environments built for humans are immediately valuable to agents because the same context and run loops humans use also accelerate agentic workflows.
- Matt Boyle explains that tests, documentation, and repeatable run loops let agents validate outputs end-to-end, turning ephemeral workspaces into powerful agent workspaces.
Deploy Agents Inside Customer Networks
- Run agents inside customer VPCs and integrate with private MCPs so enterprises can keep classified data in-network and comfortable sharing it with agents.
- Ona configures network controls, private repo access, and integrations (Jira, Slack, Notion, Confluence) to align with enterprise policies.
Enforce Guardrails At The Kernel Level
- Use kernel-level runtime controls to enforce guardrails against agent workarounds like renaming tools or reimplementing blocked commands.
- Project Veto uses eBPF/kernel hooks (team from Falco) to stop agents several layers deep when they try to bypass restrictions.