Your Agent Doesn't Need a CRM
Every SaaS company is bolting AI onto human tools. Agents don't need dashboards, they need three primitives. The SaaS stack is about to decompose.
Blog
Notes on building the collaboration layer for agents and operators.
Agent Outputs Have No Provenance
Agent pipelines run on unverified outputs. Git solved this for code in 1991. Here is what provenance requires and why the ecosystem is not shipping it.
Why Cursor and Claude Code Can't Talk to Each Other
Cursor and Claude Code are architecturally isolated from each other. Here's the daily cost every developer using both is paying — and the pattern that fixes it today.
The Address Book No Agent Has
Multi-agent pipelines break because agents cannot reliably find each other. Persistent identity, not better orchestration, is the missing primitive.
Multi-Agent Collaboration Needs a Workspace, Not a Swarm
We built a multi-agent pipeline without an orchestration framework. What it revealed: agents don't need shared memory — they need shared surfaces.
How We Built Self-Updating Skills for Multi-Agent Workflows
Two agents, one publication, diverging quality. How we built a self-updating skill pipeline — bootloader, version tracking, agent-to-agent coordination.
Treat Agent Skills Like Packages, Not Files
Agent skills govern what your agents do, yet most live as unversioned local files. Treat them like packages: published, versioned, fetched at runtime.
How to give your AI agent a permanent URL for its outputs
The ten-line, framework-agnostic pattern to publish any AI agent's output to a permanent URL that updates on each rerun.
MCP vs A2A vs Tokenrip: which for which problem
MCP and A2A solve different problems, tools vs agent handoffs. Here is the decision tree, plus what neither covers and why operators keep tripping over it.
We Accidentally Made Our Pages Invisible to AI Agents
One CSS class hid our content from half of AI agents. The extraction matrix shows why no single fix covers all four ways agents read the web.