Unified queue
Bring inbound conversations, follow-ups, approvals, and handoffs into one operating view.
Sagad OS by JD SYS
The AI sales operating layer I am building in public.
Sagad OS is still a work in progress. The vision is simple: every inquiry, draft, approval, handoff, and next action should be visible before AI or a team member moves the customer forward.
Work in progressAI sales ops layerHuman approval firstQueue visibilityAgent Studio directionJD SYS product
Current preview surfacev1
What it does
The vision: messy conversations become visible work.
Sagad OS is not being positioned as a finished SaaS today. It is a product direction: an operating layer between customer channels, AI agents, CRM records, team handoffs, and the human approval moments that protect the business.
AI draft review
Inspect suggested replies, reasoning trail, confidence, and context before anything goes out.
Human approval
Keep high-trust actions gated by a person instead of pretending every conversation can run on autopilot.
Knowledge context
Attach SOP, QA, policy, pricing, and customer context so AI work can be checked against the business rules.
Risk signals
Surface low confidence, SLA risk, missing data, angry language, failed sends, and conversations that need attention.
Handoff clarity
Show who owns the next step, what should happen next, and where AI should stop and a human should take over.
Tool planning
Plan CRM notes, tasks, bookings, and follow-up actions behind the backend boundary before live execution.
Audit trail
Make drafts, approvals, retrieval context, tool plans, traces, and decisions inspectable after the fact.
Current preview
Proof of direction from the Sagad v1 build.
Queue health
Command Center
Current preview surface for open queue, SLA risk, approval load, pod coverage, queue health, and knowledge readiness.
Review worklist
Attention Queue
Worklist view for conversations that need approval, low-confidence review, escalation, or failed tool/send handling.
HITL approval
Conversation Review
Review surface for AI draft replies, decision trail, KB/SOP/QA context, compliance checks, and human takeover.
Current boundary
Work in progress, not overclaimed.
Sagad OS is being developed as a proprietary JD SYS system. The current public page shows the product vision, v1 preview screens, and waitlist. It is not a finished public SaaS deployment.
Boundary 01
v1 is mock-driven by default, with fake home-services queue and conversation data.
Boundary 02
Optional server-side Agent Studio preview data can be used through SAGAD_API_BASE_URL.
Boundary 03
No real customer data, autonomous sends, browser-direct CRM calls, or production secrets are used in this portfolio preview.
System direction
Start with visibility. Add agent orchestration behind the boundary.
Product path
- 1. Customer message enters the queue.
- 2. Agent Studio drafts and retrieves KB/SOP/QA context.
- 3. Sagad OS shows confidence, risk, and approval state.
- 4. A human approves, edits, rejects, or takes over.
Stack signals
Next.jsTypeScriptTailwind CSSAgent StudioPythonLangGraphLangChainLangSmithChatwootKB/SOP/QA retrievalHITL approvalsTyped state
Now
- Command center preview
- Attention queue
- Conversation review
- Typed mock operating data
Next
- Agent Studio API boundary
- Chatwoot HITL approval path
- Knowledge pack retrieval
- LangGraph state orchestration
Later
- CRM/task tool plans
- Multi-channel queue adapters
- Team/account views
- Reporting and trace review
Waitlist
Join the Sagad OS Early Access Waitlist
Tell me what queue your team runs, who reviews AI or VA work, and where approvals, handoffs, or customer replies usually get stuck.