The problem and intended users
Support teams need to connect incoming requests with trusted knowledge, while retaining control over actions that could affect a customer or external system. CxOps-ai explores that boundary as an applied customer-experience engineering project.
What the current project describes
The repository documents ticket analysis, retrieval-augmented knowledge, action selection, human approval, Zendesk integration and execution audit trails.
Architecture and documented contribution
The documented stack includes a Next.js/React interface, FastAPI and SQLAlchemy services, PostgreSQL and pgvector. Its central approach separates model reasoning from deterministic authorization: a recommendation is not itself permission to execute. The README names Zaker Hussain Rasooli as author of this applied AI engineering project.
Limitations and lessons
The project is described as actively developed. Broader evaluation datasets, additional integrations and expanded tracing appear in its stated future work. Interface counters are not presented here as customer outcomes or independently audited performance. This case covers the current CxOps-ai project, not the planned RISPU website chatbot.