QDNAAdvisory on LLM inference and training platforms, on-premises or hybrid

Tutorial: Operating a Digital Organisational Twin (DOTS)

This tutorial demonstrates how to operate a Digital Organisational Twin (DOTS) to orchestrate specialised AI agents within a secure, sovereign on-premises environment.

By QDNA · Published on · Visual Tutorial Format (80% visuals, 20% text)

Short answer. A Digital Organisational Twin (DOTS) provides a real-time software replica of the enterprise. In sovereign on-premises mode, our web interface oversees the hierarchy, monitors agent heartbeats, arbitrates HTTP 409 quarantines and enforces strict financial quotas.

Step 1: Map the organisation with the organisational twin

The organisation is mapped by assigning scoped managerial entities to each specialised team of autonomous agents. This hierarchy establishes clear accountability across operations without exposing any personal employee identifiers.

Digital Organisational Twin web interface: organizational chart and managerial entities.
Figure 1: Multi-level hierarchy and delegation boundaries within the Digital Organisational Twin.

Step 2: Supervise live agent activity and execution

Live agent activity is continuously monitored through heartbeat telemetry and real-time execution status streams. The supervision cockpit detects latency spikes and idling workers before disruptions impact production.

Operational control tower: live heartbeat monitoring and agent status.
Figure 2: Operational control tower: agent heartbeat state, idle economy and live event log.

Step 3: Arbitrate HTTP 409 quarantine and approvals

The HTTP 409 quarantine gate intercepts every action exceeding financial, technical or compliance thresholds. Operations are safely paused until a verified human operator signs the approval token.

Collaborator console: HTTP 409 quarantine arbitrations and human-in-the-loop validation.
Figure 3: Collaborator console: HTTP 409 quarantine tickets and human approval buttons.

Step 4: Isolate ReBAC memory and knowledge graphs

Knowledge graphs and episodic memory are partitioned across business units using role-based and relationship-based access controls. Sensitive compensation records remain masked while non-confidential operational facts synchronize across teams.

Partitioned semantic graph and memory: ReBAC isolation and blind aggregation.
Figure 4: Segmented knowledge graph (KGM): ReBAC policies, NDA isolation and blind HR projection.

Step 5: Trace agent execution with OpenTelemetry

Agent execution traces are propagated using W3C TraceContext headers and OpenTelemetry spans. The waterfall visualizer connects high-level business goals down to low-level tool calls and verification steps.

OpenTelemetry tracing: W3C TraceContext spans and causal waterfall.
Figure 5: OpenTelemetry telemetry waterfall: W3C TraceContext spans and agent execution latency.

Step 6: Enforce financial quotas and token ceilings

Financial quotas and hard token budgets are enforced per agent to prevent inference cost overruns. Once a threshold is reached, inference calls are automatically blocked until administrative review.

Fleet financial quotas: monthly budgets and hard spending rules.
Figure 6: FinOps dashboard: monthly fleet budget ($2,420 / 478M tokens) and hard blocking rules.

Comparison: Unregulated Swarms vs Sovereign DOTS

Deploying a sovereign organisational twin replaces unconstrained agent swarms with deterministic accountability, strict audit trails and guaranteed data containment on-premises.

Governance dimension Unregulated agent swarms Sovereign DOTS on-premise
Role boundariesIsolated agents without clear hierarchyMulti-level organigram and strict accountability
Critical actionsUnchecked direct executionHTTP 409 quarantine and mandatory human signature
Memory & contextFlat shared memory leaking confidential dataDual topology with blind aggregation and ReBAC
Compute auditabilityDispersed unindexed logsOpenTelemetry waterfall and local telemetry daemon
Financial governanceUncapped API billing with unexpected spikesGranular quotas and 100% hard blocking rules

Frequently asked questions

What is a Digital Organisational Twin (DOTS)?

A Digital Organisational Twin (or Digital Organisational Twins) is the real-time software replica of an organisation, modeling its roles, processes and decision flows.

Why run an organisational twin in on-premises DOTS mode?

An on-premises deployment guarantees data sovereignty, regulatory compliance and prevents sensitive business intelligence leaks to external providers.

How does the HTTP 409 quarantine gate protect operations?

The HTTP 409 quarantine mechanism automatically halts high-risk actions exceeding predefined thresholds, requiring human operator sign-off before execution.

Deploy your Digital Organisational Twin with QDNA

Consult with our architects to deploy your Digital Organisational Twins and operational cockpit in a sovereign, on-premises DOTS architecture.

Book a call

Further reading

References