Blog

Engineering deep-dives, roadmap updates, and security research from the Tropic team.

AgentsSecurityEngineering2026-04-17

The Three Columns Behind the Ten Walls

Ajey Gore named the walls. They fall into three columns: Infrastructure, Authority, Influence. That's the framing I've been building Tropic around.

EngineeringAgents2026-04-15

LOD Memory: Why Agent Memory Is Broken and What Terrain Generation Taught Me About Fixing It

Flat memory architectures force permanent decisions about what matters, and those decisions prove wrong over time. A concept borrowed from game engines: keep everything, return it at dynamic resolution based on how relevant it is right now.

SecurityAgents2026-04-11

Two AI Cybersecurity Products Walk Into a Bar

Anthropic's Mythos and OpenAI's equivalent find vulnerabilities in code at rest. Neither governs the agents already running in production. Here's the gap.

EngineeringSecurity2026-03-19

Setting Up NemoClaw Step-by-Step

A practical walkthrough of installing and configuring NVIDIA's NemoClaw on a Linux VM, from API key to sandbox to first chat.

EngineeringSecurity2026-03-18

How Tropic Complements NemoClaw

NVIDIA's NemoClaw brings kernel-level sandboxing to OpenClaw agents. Tropic wraps it with infrastructure, credentials, audit logs, and a dashboard. Here's how they fit together.

SecurityEngineering2026-03-15

What a Security Control Plane Needs to Do

How Tropic implements the four pillars of MDDI's Model AI Governance Framework for Agentic AI: Bound, Accountable, Control, and Transparent.

OpenClaw SeriesSecurity2026-03-14

What is a Port?

The first entry in our OpenClaw Series. A plain-language guide to ports — what they are, why they matter, and why so many OpenClaw instances got hacked.

RoadmapSecurityAgents2026-03-05

Tropic Roadmap: Where We're Headed

Our vision for the future of Tropic: agent observability, expanded integrations, chaos testing, security hardening, and more.