Ideas on
building with AI.
Thoughts on multi-agent systems, parallel execution, and the future of building with AI.
The Self-Driving Company
Everyone is starting to talk about self-driving companies. The conversation is not about removing people, it is about agents carrying execution while people keep direction. We think the unsolved piece is trust.
Why Agent Safety Needs to Live Outside the Model
Most AI agent products enforce safety through the system prompt. Research shows these defenses can be bypassed more than 78% of the time. Mode Agent enforces boundaries at the orchestration layer, independent of any model's inference.
5 min readWhy multi-model parallel agents are the future of AI-assisted work
Running agents in parallel is just the beginning. The real unlock is routing different agents to different models based on what each task actually needs.
7 min readSandboxed by default. How we built agent permission boundaries
AI agents that stop and ask permission every ten seconds aren't useful. Here's how Mode Agent gives agents full autonomy while letting you set the boundaries that matter.
6 min readWhy we chose 500+ models over one great model
The best model for planning is different from the best model for coding or testing. Provider-agnostic routing lets each agent role use what it's actually best at.
7 min readGet new essays in your inbox
Multi-agent systems, parallel execution, and building with AI. No spam, unsubscribe anytime.