QingZoneX Engineering

Building QTable: Engineering an AI-Native Work System

This is not a stream of product announcements. It is a series about real engineering questions: why QTable exists, how AI can safely operate business data, how context is built, how agents execute work, and what it takes to make a project genuinely open source.

Series10 / 10

Self-hostedRecoveryObservability

Self-hosted Is More Than docker compose up: How QTable Turns Upgrades, Backup, Recovery, and Observability into System Boundaries

The hard part of self-hosting is not starting containers; it is upgrading, recovering, and operating the system for years. This article follows QTable’s PostgreSQL / Redis / MinIO / QTable / QTableUI Compose stack, Alembic migrations, health checks, release gates, ChangeSets, and runtime traces to separate today’s operational foundations from the backup/restore, readiness, metrics, and alerting work still ahead.

September 28, 2026Read article →

AutomationDeterministic RuntimeQTable

An Agent Should Not Run Every Automation: Why QTable Keeps Deterministic Execution in the Automation Engine

AI-native does not mean every workflow belongs to a model. This article follows QTable’s Rule / Event / Execution model, ChangeSet event materialization, scheduler, worker, run-as permissions, retries, rule versions, loop guard, and execution history to explain how Agent and Automation responsibilities should differ.

September 28, 2026Read article →

Supply ChainSBOMProvenance

Open Source Is More Than Making a Repository Public: How QTable Builds Supply-Chain Trust with CI, SBOMs, and Provenance

Publishing source is only the first step; users deploy dependencies, images, and release artifacts. This article follows QTable / QTableUI CI, dependency audits, CycloneDX SBOMs, Trivy, BuildKit provenance, OCI metadata, image digests, and full-stack release gates to connect reviewed source to verifiable artifacts.

September 28, 2026Read article →

AuthorizationAgent SecurityQTable

An Agent Is Not a System Administrator: Why QTable Keeps Agent Permissions Below the Current User

An Agent may read tables, analyze work, and call Tools, but being AI must not grant more authority than the current user. This article follows QTable’s current identity flow, Workspace / Item permissions, row permissions, Skill authorization, relation sanitization, and Action Plan Apply revalidation.

September 27, 2026Read article →

View ModelData ConsistencyQTable

Grid, Kanban, and Gantt Should Not Store Three Copies of the Data: Why a QTable View Is Only a Projection

The same task is a row in Grid, a card in Kanban, and a bar in Gantt. This article follows QTable’s current Record / View model and QTableUI implementation to separate business state from view state, including Kanban ordering, Gantt field mapping, realtime convergence, and current consistency gaps.

September 27, 2026Read article →

Tool ChainAgent RuntimeQTable

An Agent Is Not a Sequence of Function Calls: How QTable Turns Tools into a Runtime

Real agent work often means task splitting, estimation, Schema inspection, Gantt generation, and Record creation. This article follows QTable’s current Multi Tool Chain Runtime through ToolChainPlan, dependencies, data passing, retry / confirmation / rollback, observability, and the DAG scheduling work that is still incomplete.

September 27, 2026Read article →

Context EngineAgent RuntimeQTable

Context Is Not More Prompt Stuffing: How QTable Builds Real Business Context for Agents

Once an agent enters a business system, the hard part is not how much the model can remember. This article follows QTable’s current Context Engine through scope, permission-aware Table context, conversation compression, Cache / Session / Snapshot, and runtime injection.

September 27, 2026Read article →

AI safetyAction PlanQTable

Why AI Writes Should Not Go Straight to the Database: QTable's Preview → Confirm → Apply

Once AI can change business data, stale, unauthorized, or misunderstood decisions can become durable facts. This article follows QTable’s current Action Plan, record-version checks, permission revalidation, and ChangeSet audit path to explain why Preview → Confirm → Apply is an execution boundary.

September 27, 2026Read article →

Data modelAgentsQTable

Why Multidimensional Tables Make a Good Business Data Foundation for AI Agents

Once an agent enters a real business system, fluent generation is not the hard part. This article starts from QTable’s current Field / Record / View / Relation / Permission model and asks why multidimensional tables make a practical layer between natural-language intent and executable work.

September 16, 2026Read article →

Product thinkingAI NativeQTable

Why Build Another Open-Source Multidimensional Table? QTable Is Not Really About Tables

Airtable, Notion, and modern multidimensional-table products are already mature. Why build QTable? The first article explains the product thesis behind an AI-native structured work foundation.

September 13, 2026Read article →

The canonical Markdown source lives in this repository. Syndicated copies on external engineering communities should point back here.