Operations

System jobs

System jobs are Qorpera’s autonomous background work. They come in two categories — infrastructure jobs that keep the platform’s understanding current, and delegated operational jobs the operator has explicitly chosen to run without per-instance approval.

Infrastructure jobs

These run regardless of operator action. They are the “background intelligence” that keeps the wiki accurate and the reasoning engine calibrated:

Onboarding analysis. The multi-phase discovery pipeline that runs during Research.
Document intelligence. Deep analysis of uploaded documents — classification, expertise assembly, full-document comprehension, cross-document correlation, analytical wiki synthesis.
Research synthesis. The corpus pipeline that builds system intelligence from professional literature.
Wiki quality monitoring. Independent verification of page claims, trust-level adjustment, quarantine enforcement.
Heuristic consolidation. Raw learnings on resolved situations distilled into playbook heuristics; weekly Opus meta-reasoning proposes new heuristics and flags contradictions.
Detection audit. Pre-filter accuracy compared against full LLM detection; regeneration when drift is detected.

Delegated operational jobs

Each situation type you delegate becomes a system job. The job detects instances of that type, runs the full reasoning cycle, and executes the action plan — without surfacing for human approval.

System jobs respect the same governance as supervised reasoning. Policy rules are evaluated before every action; high-stakes actions can still be configured to require approval even within a delegated type.

Telemetry and revocation

System jobs log every decision to the situation wiki page — same trail as supervised situations. You can audit any executed action, compare to what the supervised version would have done, and revoke autonomy for that situation type at any time. A revoked type reverts to propose-mode immediately; in-flight instances complete under the original disposition.

Admin visibility

System jobs are visible to admins under /system-jobs. Members don’t see the system-jobs surface — the delegation decision is an operator-level one.

Previous← Projects NextConnecting tools →

Infrastructure jobs

These run regardless of operator action. They are the “background intelligence” that keeps the wiki accurate and the reasoning engine calibrated:

Onboarding analysis. The multi-phase discovery pipeline that runs during Research.

Document intelligence. Deep analysis of uploaded documents — classification, expertise assembly, full-document comprehension, cross-document correlation, analytical wiki synthesis.

Research synthesis. The corpus pipeline that builds system intelligence from professional literature.

Wiki quality monitoring. Independent verification of page claims, trust-level adjustment, quarantine enforcement.

Heuristic consolidation. Raw learnings on resolved situations distilled into playbook heuristics; weekly Opus meta-reasoning proposes new heuristics and flags contradictions.

Detection audit. Pre-filter accuracy compared against full LLM detection; regeneration when drift is detected.

Delegated operational jobs

Each situation type you delegate becomes a system job. The job detects instances of that type, runs the full reasoning cycle, and executes the action plan — without surfacing for human approval.

System jobs respect the same governance as supervised reasoning. Policy rules are evaluated before every action; high-stakes actions can still be configured to require approval even within a delegated type.

Telemetry and revocation