ProjectLoom is a proof-of-concept AI workflow demo. It shows how an agent can read an incoming
operations email, classify the issue, retrieve relevant context, recommend a plan, and prepare
a reviewed handoff package for existing business systems.
- Generate EmailCreate a synthetic production email for the demo queue.
- Run WorkflowStream classification, routing, evidence retrieval, and recommendation drafting.
- Approve / Modify / RejectHuman review controls whether a handoff package is prepared.
- Send HandoffSimulate emails, tickets, tasks, logs, and audit events in the demo environment.
Proof-of-concept disclaimer
This is not production software. It is intentionally limited, may not be optimized for speed,
and should not be interpreted as the final product experience. A production version would be
designed around a customer's existing systems, security requirements, approval policies,
integrations, and operational workflows.
Data boundary
All emails, SOPs, incidents, tickets, handoffs, and system events in this demo are fictional,
mocked, or synthetic. No private company data is used. Public and synthetic textile-manufacturing
datasets are used only to make the scenarios feel closer to day-to-day plant operations.
Built with
- Python
- FastAPI
- Pydantic
- Uvicorn
- OpenAI API configured by environment
- HTML
- CSS
- Vanilla JavaScript
- Render deployment
- JSON/Markdown knowledge base
- Public/synthetic CSV datasets
The final handoff sequence is simulated for the proof of concept. It does not connect to or update
real inboxes, ticketing systems, dashboards, ERP/MES tools, or private plant systems.