T2.5 · DispatcherCross-Module⭐ Triple Impact

Evidence Loop — WhatsApp Field Verification

Challenge the AI's analysis. Get field confirmation in minutes.
Improve Efficiency📈 Increase Revenue💰 Lower Expense
Directed Acyclic Graph (DAG) EngineConfidence Propagation AlgorithmMulti-Agent Debate ProtocolImmutable Audit Trail

📖 The Scene

When Siti questions a recommendation, she asks NayaCore for proof. NayaCore reveals its full evidence chain in an interactive DAG format, merging raw telemetry with field reports and physics laws.

👤

Siti + Budi

Operator + Field Worker

Evidence Chain Visualization

AI trust is built on transparency. NayaCore doesn't just give answers; it shows every step of its reasoning — from raw signals to human evidence and AI exclusion logic.

🔍
09:38 WIB

Requesting AI Proof

When presented with a high-stakes recommendation, Siti asks NayaCore to prove its logic. NayaCore responds with a structured breakdown of the 4 supporting pieces of evidence and 3 excluded hypotheses.

⚖️AI Evidence Query
Naya, tunjukkan bukti untuk INC-130809:38
NayaCore🔍 Evidence Chain — INC-1308
Bearing Inner Race Degradation
Press #3 · Confidence: 87%

━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📡 SENSOR (3 bukti):
1. Vibration 9.2 mm/s (batas: 8.0) ✅
2. FFT puncak 48Hz = frek bearing ✅
3. Trend +0.30 mm/s per hari ✅

📱 MANUSIA (1 bukti):
4. Budi: "bearing kering, sisi kanan bawah" ✅

🔀 ELIMINASI:
❌ Bukan imbalance (load konstan)
❌ Bukan coupling (Budi bilang 'bearing')
❌ Bukan fondasi (progressif)

🧠 KESIMPULAN AI:
Inner race degradation. Penyebab: kekurangan pelumasan.
Lihat grafik lengkapapp.naya.bireka.id/millos/insights/1308/evidence
09:38
09:39 WIB

Interactive Interrogation

Siti challenges the AI, asking why it ruled out a coupling issue. NayaCore defends its reasoning using both physics (FFT frequency mismatch) and NLP (cross-referencing Budi's location report).

🧠AI Interrogation
Kenapa bukan coupling?09:39
NayaCore🔍 Alasan Eksklusi: Coupling

Coupling misalignment biasanya menunjukkan pola 2× RPM (49Hz) di FFT, bukan 1× BPFI (47.3Hz).

Selain itu, Budi secara spesifik menyebut "bearing" dan "sisi kanan bawah" — coupling berada di sisi kiri (dekat motor).

Confidence bahwa ini BUKAN coupling: 95%
09:39
🕸️
09:42 WIB

PWA Graph & Audit Trail

If Siti opens the PWA, she sees a fully interactive Directed Acyclic Graph (DAG) of the evidence chain. Every node is clickable, showing sensor health, algorithm confidence, and an immutable audit trail for compliance.

EVIDENCE DAG
Vibration: 9.2 mm/s
PASS
FFT Peak: 48Hz
PASS
Human: "bearing kering"
PASS
CONCLUSION: Inner Race Degradation (87%)

📊 Before vs. After

Traditional Operations vs. NayaCore

MetricTraditionalNayaCore
Operator trust in AI recommendations~40% (black box anxiety)~85% (transparent reasoning)
Time to validate AI recommendation15–30 min (manual investigation)< 2 min (click through evidence nodes)
Dispute resolution timeHours (he-said-she-said)Minutes (shows divergence point)