Case Studies
Production deployments. Real outcomes.
Where Opstral runs in production today, what it replaced, and what changed. Customer identities are withheld by request.
TradeChef: 1,500 Instruments, Every Stop-Loss on Time
An options trading platform put its live tick-to-order chain, nightly bhavcopy pipeline and multi-year backtesting under one latency budget: leading-indicator detection, guilty-hop isolation in 90 seconds, blast radius priced in rupees, and Sherlock-verified recovery.
- 150 mstick-to-order budget, enforced
- 90 secto guilty-hop isolation
- 3–4 minearly warning before breach
Hardware-Depth Observability for a Tier-1 North American Telco
Thousands of servers running heavy data workloads, with every HPE iLO interface a disconnected island, unified into one governed layer, down to temperature and component health, with Sentinel acting before failures land.
- ~70%issues caught before impact
- ~40%faster MTTR
- 100%iLO endpoints centralized
Carrier-Scale Network Observability: 27,000+ Devices, One Governed NOC
Fault and performance data from twenty-seven thousand routers, switches and devices, over SNMP, Kafka and polling, unified into near-real-time AI investigation, conversational live values, MOP automation and governed incident creation.
- 27,000+devices in one pane
- 43%MTTR reduction, measured
- 85%routine ops automated
Governed Day-2 Operations Across 2,000+ Nodes with ProcBot
Health checks, log capture with AI analysis, SSH under RBAC, and mass patching that runs as governed MOPs with precheck, postcheck, audit and automated verification, live on two thousand nodes, forecast toward one hundred thousand.
- 2,000+nodes live today
- ~100knodes forecast
- ~70%faster patch rollout
Self-Healing Network Operations: The Loop That Never Skips a Step
Symptoms discovered across seven network data domains, root causes diagnosed through RCA trees built from rules, ML and knowledge docs, remediation executed under governance, and every loop closed on quantitative feedback, with full provenance.
- ~60%faults resolved end to end
- ~50%faster root cause
- ~35%fewer repeat incidents
Central Agent Orchestration Across Multi-Provider Managed-Services Estates
A global consumer-goods enterprise, delivered with a Big Four consulting partner: Opstral sat above every managed-services provider's own agents, coordinating them agent-to-agent for RCA, resolution and automated postmortems, triggered by incidents or Copilot chat.
- A2Aprovider agents conducted
- Self-builtagents in Agent Studio
- ServiceNowITSM actions via connector
Agent-Governed Release Automation: Hundreds of Daily Builds
An AI agent fronts the whole release process: email intake with Git URL and tags, a four-stage approval chain it manages itself, Jenkins pipelines triggered and monitored step-by-step, exact-fix feedback emailed to developers, and GitHub PRs raised with its review attached.
- 100sdaily build iterations
- 2engineers run the estate
- 20+environments managed
More production case studies in progress