Approve the source
A source enters scope only after explicit allowlisting and connector-policy binding.
LOCAL-FIRST OPERATIONAL INTELLIGENCE
Qavirio NERVE
Understand how business processes actually run — and trace every signal back to evidence.
NERVE 1.0 Final reads explicitly approved business sources in read-only mode, normalises events, learns process routes and timing patterns and presents evidence-backed signals to analysts and owners. Version 1.0 is deliberately not designed to write back automatically to source systems.
01 · Purpose
Why this product exists.
NERVE 1.0 Final reads explicitly approved business sources in read-only mode, normalises events, learns process routes and timing patterns and presents evidence-backed signals to analysts and owners. Version 1.0 is deliberately not designed to write back automatically to source systems.
02 · Operational problem
Many organisations have process data but limited controlled insight into what those processes actually do.
Events are spread across files and systems. Abnormal timings or routes are noticed late, and a signal without evidence is difficult to trust. NERVE makes approved events locally analysable without adding write capability to the source.
Build process routes and timing from observed events rather than assumptions.Link signals to concrete event references and source context.Separate analysis from authority to act.Keep evidence local and export it in a reviewable form.
03 · How it works01 02 03 04 05 06
From approved source to evidence-backed process insight.
Read-only ingestion
CSV, XLSX or controlled folder connectors read approved input without a write path back to the source system.
Normalise events
Raw input is converted into consistent business events with source and timing context.
Business Graph & Process Learning
Entities, transitions, routes, support and timing baselines are built from local evidence.
Signals & Evidence
A detector produces a signal with confidence; Evidence Chain 2.0 links signal, evidence, process model, policy and conclusion.
Human review
NERVE supports investigation and decision-making; ACT is off in 1.0 and source-system write privileges are zero.
04 · Capabilities
Business Graph
Visualise observed entities, relationships and process transitions from normalised events.
Process Learning
Learn routes, timing baselines, support and anomalies from repeatedly observed evidence.
Evidence Chain 2.0
Signal → Evidence → Process Model → Policy → Conclusion as an explicit explanation of the current conclusion.
Security Console
RBAC self-test, database health, ingestion kill switch, security report, backup and Trust Pack export.
Verified backup
SQLite backup is staged locally, verified and returned with integrity evidence.
Audit & Trust Pack
Hash-linked audit and an exportable Trust Pack make status, policy and evidence inspectable.
05 · User experience & roles
Business analyst / process owner
Investigates routes, timing, signals and evidence without source-system write capability.
Auditor
Reviews audit chain, evidence, policy and Trust Pack without operating the runtime.
IT / Administrator
Manages Windows identity/RBAC, approved sources, connectors, backup and deployment.
Security team
Reviews read-only boundary, audit integrity, source policy, database health and external gates.
06 · Use casesPROCESS TIMING ROUTE DEVIATION AUDIT
A normal route suddenly takes much longer
Process Learning compares observed transitions and timing with prior evidence; the reviewer can trace the signal back to concrete events.
A process follows an unusual sequence
NERVE shows the observed route, support/reliability where available and the event context behind the deviation.
A reviewer needs to know why a signal existed
Evidence Chain 2.0 and the audit chain provide a bounded, inspectable explanation without presenting the signal as black-box truth.
Real product interface
Process Learning, Signals & Evidence and Security Console.
IN-DEPTH EXPLANATION
NERVE in detail: from approved source to explainable operational signal.
NERVE is built for organisations that have a great deal of operational data but cannot automatically see how work really moves through their systems. A formal process model describes how a process is intended to work; event data shows what actually happened. NERVE makes that second reality visible in a controlled way. It reads only explicitly approved sources, normalises relevant occurrences into a common event layer and uses that layer to examine process routes, timing and deviations. The aim is not to generate as many alerts as possible, but to trace each signal back to concrete evidence so an accountable person can judge whether something genuinely requires attention.
Read-only is an architectural choice
The 1.0 product line deliberately separates understanding from execution. Approved connectors ingest or read data without creating a general write path back into source systems. That reduces the blast radius of the intelligence layer and makes the trust model easier to reason about: NERVE can learn from a workflow without silently changing the workflow it observes. The organisation remains responsible for the operational decision, while the software provides structured evidence, timing context and process knowledge. If a later deployment ever introduces action capabilities, those capabilities must be treated as a separate authority plane rather than an incidental extension of analytics.
From events to process understanding
Normalisation allows data from different approved sources to become comparable. A useful event needs enough structure to answer what happened, when it happened, which object or process it concerned and where the evidence came from. Once events share that language, NERVE can reconstruct recurring paths and timing patterns rather than relying only on hand-written process maps. Process Learning is therefore descriptive before it is predictive: the software first establishes a measurable baseline of observed behaviour, then identifies cases that depart from that baseline. A slow case, an unusual route or a missing expected transition becomes meaningful because it is compared with previously observed process evidence.
Signals remain reviewable
A signal without evidence quickly becomes another opaque score. NERVE is designed so a user can move from the signal back toward the events and source context that produced it. This is especially important when operational data is incomplete or ambiguous. The software should be able to say that a pattern is unusual without pretending to know the business reason automatically. That distinction supports investigation: an analyst can confirm a real bottleneck, identify a data-quality issue or conclude that an apparent deviation is actually legitimate. Audit, source control and evidence export make that reasoning inspectable rather than ephemeral.
07 · Architecture
Local evidence path and a separate security plane.
NERVE 1.0 Final is a Windows x64/.NET 10/WPF desktop product with local SQLite evidence storage. The runtime path is approved read-only sources/connectors → hardened ingestion → normalised business events → local evidence store → Business Graph/Process Learning → evidence-backed signals. A separate security plane governs Windows identity, RBAC, source policy, connector policy, ingestion kill switch and hash-linked audit.
Approved sources→Read-only ingestion→Local SQLite evidence→Business Graph / Learning→Signals / Evidence
08 · Security & data
THINK / READ-ONLY · ACT OFF · EXTERNAL AI OFF · WRITE PRIVILEGES 0Explicit source allowlist and connector/path bindingWindows identity and role bindingGlobal ingestion kill switch while preserving existing evidenceHash-linked audit is tamper-evident, not cryptographically immutable against privileged rewriteNo keylogging, screenshots, browser history, camera/microphone or hidden employee timer
09 · Integration
1.0 supports controlled file connectors, not unrestricted system integration.
The current connector family includes CSV, XLSX and Folder. Sources are explicitly approved and connectors are bound to a source and path. NERVE 1.0 has no central SSO/OIDC and no configured external AI provider. Future integrations must be designed and qualified separately.
10 · Deployment & hosting
Windows workstation / VM
Per-machine installation under Program Files; runtime operates without elevation after installation.
Enterprise Lockdown
Windows groups can be bound to NERVE roles; managed endpoints and full-disk encryption are recommended for enterprise deployment.
Local data
Evidence, logs and configuration remain local in the Windows environment unless the customer deliberately chooses another controlled setup.
11 · Operations & support
12 · Product status
1.0 FINAL · READ-ONLY BASELINE
The local 1.0 Final product baseline is internally complete. Broad external enterprise rollout remains subject to production-specific gates such as commercial code signing, independent penetration testing and privacy/retention review.
Commercial Authenticode/code signing for broad external rolloutIndependent penetration test / relevant security reviewCustomer-specific privacy, DPIA, retention and governance review where requiredOrganisational assurance such as ISO/SOC is not implied by the software build
13 · Contact / demo
Request technical presentation →Review NERVE with business and IT/security in the same conversation.
A technical presentation can show Process Learning, Evidence Chain 2.0, read-only boundaries, Windows RBAC, backup/audit and deployment gates without opening a public interactive demo.