Скачать или смотреть Daxi: 3. High Stakes Workflows

daxi: High-Stakes Workflows — Episode 3

When decisions carry real cost, delay, or safety risk, you can’t rely on ad-hoc scripts or single-shot chat flows. In this episode, we show how daxi (Domain-Specific, Event-Driven DSL for Actions with eXecutable Insights) runs high-stakes manufacturing workflows with human oversight and AI working together.

What you’ll learn

1. The High-Stakes Problem
Why quality, yield, and safety decisions fail when processes aren’t traceable, repeatable, or auditable.

2. Daxi’s Solution
An event-driven DSL that orchestrates end-to-end flows (data retrieval → analysis → HITL approval → reporting) across services like vector search, ML/SPC, and LLM helpers—while keeping state, retries, and audit trails.

3. “95% Proof”
How codifying domain context (4M change & NG issue patterns) covers the majority of real cases and reduces time-to-decision with consistent evidence.

4. Empowering Experts & AI
Human-in-the-loop (HITL) gates, explainable alerts, and judge/evidence scoring so engineers stay in control while AI accelerates the heavy lifting.

How it works (at a glance)

1. Event-Driven Orchestration: Each step emits/consumes events on a queue for reliability and scale.
2. Domain Context First: Structured context packages (lots, time windows, specs, linked docs) drive retrieval and analysis.
3. Analytics You Can Trust: Built-in SPC/ML (e.g., CUSUM/EWMA, pre/post comparison) to detect shifts, drifts, and effectiveness after actions.
4. HITL by Design: Formal approval tasks with pause/resume, comments, and traceability.
5. Final Reports: Auto-generated, evidence-linked summaries you can file to QMS.

---
🔔 If you’re interested in domain-specific DSLs, event-driven AI workflows, and real factory use cases, please subscribe.
📺 Watch the episode series in this playlist:    • daxi  

About daxi
daxi is an Event-Driven DSL AI Workflow platform for manufacturing.
With a Kafka-based event architecture and a declarative DSL, it automates complex shop-floor workflows and enables evidence-based decision-making through collaboration between AI and humans (Human-in-the-Loop, HITL).
👉 Domain-Specific, Event-Driven DSL for Actions with eXecutable Insights
👉 “DSL for Actions, executed by Context and Insight.”

#daxi #ManufacturingAI #EventDriven #DSL #Quality #SPC #CUSUM #EWMA #HITL #RCA #CAPA #IndustrialAI #Kafka #RAG #LLM