Shadow Observer

Passive Workflow Intelligence Engine

9 Sources$0 LLM Cost

See what your teams actually do — without asking

Passive telemetry intelligence that discovers recurring workflows, quantifies automation potential, and maps them to Claresia skills — all without surveys, interviews, or screen recording.

9 Passive Sources
9
Local telemetry files
0 LLM Cost
$0
Deterministic inference
82ms Processing
82ms
Avg pipeline latency
Avg Workflows/Person
19
Discovered automatically
Annual ROI
$17K+
Per department avg
Zero Disruption
0
No surveys or interviews

What is Shadow Observer?

Shadow Observer is a lightweight, privacy-first telemetry scanner that reads local files already on every employee's machine — SQLite databases, JSONL session logs, git history, and OS indexes. It groups these events into activity sessions, detects recurring patterns using Jaccard similarity clustering, and classifies them into named business workflows. The entire inference pipeline is deterministic: no LLM calls, no hallucination risk, no token costs. Results include automation potential scores, estimated time savings, and direct mappings to Claresia's 56 deployment-ready skills.

Traditional vs Shadow Observer

DimensionTraditional ApproachShadow Observer
Data CollectionSurveys, interviews, workshopsPassive telemetry (automatic)
Employee DisruptionHigh — pulls people from workZero — runs silently in background
Time to Insight4-6 weeks2 days
Accuracy~30% (self-reported bias)85%+ (behavioral data)
Cost Per Assessment$50K-$150K consulting$15K-$50K (automated)
CoverageSample of roles interviewedEvery employee with a device
RepeatabilityOne-time snapshotContinuous monitoring available

How It Works

Step 1

Deploy

Install lightweight Node.js scanner on employee devices. Reads existing telemetry files — no agents, no hooks, no permissions dialogs.

Step 2

Observe

Scanner reads 9 local data sources: SQLite DBs, JSONL logs, git history, OS indexes, API metadata. All processing stays on-device.

Step 3

Infer

Deterministic 4-stage pipeline: sessionization, pattern detection, classification, scoring. Zero LLM calls. 82ms average latency.

Step 4

Prescribe

Named workflows with automation scores, time savings estimates, and direct mappings to Claresia skills. Ready-to-deploy recommendations.