CXOpsCXOps is a personal exploration of agentic content operations. I built it to think through one specific question: Given advancements in consumer AI and tooling, how feasible is it to solo-build an end-to-end, autonomous, agentic content optimization platform?
What it is — and what it isn't
This is a working technical demonstration. The modules under /features and the tools under /tools/ are functional code, not mockups. You are welcome to review the documentation (also largely LLM-generated and human-edited) at docs.cxops.cloud.
This isn't a service for sale. There is no signup, no account, no pricing, no roadmap. If you'd like to use the patterns, please do. I won't share the source code — I invested lots of personal time on it, and the point is to learn. Crack open Claude Code or your favorite IDE and start on your own version.
Architecture, top to bottom
A short tour of how the pieces fit together. Specific hosts, ports, and internal service names are intentionally omitted.
Frontend. This site is statically built with Astro; copy comes from Sanity and is fetched at build time. The browser-based tools under /tools/ run entirely client-side. Reference documentation lives at docs.cxops.cloud as a separate publication.
Application services. The platform itself runs as a set of containerized services behind an identity provider that gates anything non-public. A workflow engine orchestrates long-running jobs — content briefs, generation, approval routing, publish triggers — so the agent and UI layers can stay stateless and restartable.
Data. Two stores. A structured database holds the Brand DNA, audience personas, governance rules, and generated assets. Sanity is the headless CMS for the human-edited copy that powers this site and the docs. A vector index sits alongside the structured store so the agents can retrieve voice samples, prior work, and source material semantically.
Agents. The six modules — Ambient Cortex, Brand DNA Vault, Signal Intelligence, Content Performance Loop, Governance & Compliance, and Channel Adaptation — are implemented as Claude-based agents coordinated through the workflow engine. Each owns one responsibility (Cortex generates, the Vault constrains, Signals informs, Pulse learns, Govern audits, Adapt translates) and they communicate through the shared store rather than calling each other directly. Generation isn't a single LLM call but a planned sequence with retrieval, structured tool use, and human-in-the-loop checks where they matter.
Observability. The usual: structured logs, health endpoints, error tracking, and GA4 on the public site.
Contact
The easiest way to reach me about this is by email: admin@cxops.cloud.