โ— In UAT โ€” Coming Soon BFSI-Native Agentic AI

Neluvi Sentinel
Autonomous L1/L2 IT Support

The first AI platform that doesn't just classify tickets โ€” it investigates your systems, diagnoses root causes, fixes problems, and proves it worked โ€” with a regulator-grade audit trail.

Join the Waitlist How It Works โ†“
1,855 Tests Passing
54+ API Endpoints
12 Specialist Agents
9 Phases Shipped
<5 min Target MTTR
The Problem

BFSI IT Support Is Broken

A mid-size bank handles 8,000โ€“15,000 IT tickets per month. 70% are repeats. Engineers waste hours on the same investigation steps.

๐Ÿ’ธ
$800Kโ€“$1.6M/yr Staffing
20 L1 engineers ร— $40โ€“80K. 30โ€“40% attrition. Night shifts add 40% overhead. The cost never stops growing.
โฑ๏ธ
47 Min Average Resolution
60โ€“90 min just understanding the issue โ€” grepping logs, reading past incidents. Actual fix is 15 minutes.
๐Ÿ”„
70% Repetitive Tickets
Password resets, service restarts, batch re-runs โ€” identical steps, hundreds of times per month.
๐Ÿง 
Knowledge Loss on Attrition
Resolution patterns live in people's heads. New hires need 6โ€“12 months to become effective. No institutional memory.
๐Ÿ“‹
Compliance Overhead
Every production action needs SOX/RBI audit trails. Manual maintenance is error-prone and time-consuming.
๐Ÿ”Œ
Zero Incident Correlation
Alerts in silos. Nobody connects API slow + DB locks + batch long + recent deploy into one root cause.
"Existing tools classify and route โ€” they don't investigate or resolve."
The Solution

An Autonomous L1/L2 Engineer

Not a smarter chatbot. A full autonomous resolution loop with investigation, remediation, verification, and learning.

Ticket In
Parse
Enrich
Investigate
Diagnose
HITL Gate
Remediate
Verify
RCA
Learn
๐Ÿ”
Investigate

Agents log into Splunk, ELK, databases, Kubernetes, Control-M. Search logs, check locks, inspect infra, trace batch chains, diff configs โ€” all read-only.

6 specialist investigation agents ยท Concurrent execution
โšก
Remediate

Restart services, re-trigger jobs, grant access, fix configs, run approved scripts โ€” with pre/post health checks, auto-rollback, and tiered approval gates.

5 remediators + Fix Verifier + Escalation Agent
๐Ÿง 
Learn

Auto-generate RCAs and runbooks. Embed into RAG. Track confidence per issue type. Auto-promote proven fixes from Tier 2 โ†’ Tier 1 at โ‰ฅ95% confidence.

Qdrant RAG + confidence scoring + auto-promotion
Architecture

How Sentinel Works End to End

๐Ÿ“ฅ
Ingest ServiceNow, Jira, Email, Webhook
๐Ÿ”ค
Parse LLM + heuristic classification, 8 intents
๐Ÿ”—
Enrich Deploys, changes, CMDB correlation
๐Ÿ”
Investigate 6 agents concurrent, read-only system access
๐Ÿ”€
Correlate Multi-system root cause analysis
โœ‹
Approve Tier 1/2/3 + dynamic approval routing
๐Ÿ”ง
Fix Execute with verify and auto-rollback
๐Ÿ“š
Learn RCA โ†’ Runbook โ†’ RAG knowledge base
Technology Stack
Python 3.12 FastAPI LangGraph PostgreSQL 16 Redis 7.2 Qdrant 1.9 React 18 TypeScript Tailwind 3 Docker / K8s Gemini Pro Claude Opus JWT + RBAC AES-256
SOX / RBI
Audit Trail Ready
Multi-Tenant
PostgreSQL RLS
Business Impact

The Numbers That Matter

70%
Ticket auto-resolution target
<5 min
Mean time to resolution
90%
Reduction in L1 repetitive work
$1.2M+
Annual staffing savings (mid-bank)
100%
SOX/RBI audit trail coverage
โ‰ฅ95%
Confidence threshold for auto-promotion
โ— UAT โ€” Accepting Design Partners

Be Among the First to Deploy Sentinel

Limited design partner slots available for BFSI organizations. Shape the product roadmap and get early access before general availability.

Apply as Design Partner