Strike48 Platform

Strike48 Platform is an agentic AI security and operations platform that enables organizations to build, deploy, and orchestrate AI agents for security operations, fraud detection, and IT incident response. **Core Capabilities:** The platform provides a "Prospector Studio" environment for creating custom AI agents with defined personas, curated knowledge via GraphRAG, and explicit tool access through Model Context Protocol (MCP). Agents can be pre-built or user-created, covering use cases such as SOC Level 1 alert triage, fraud detection, and incident response automation. **Data & Log Handling:** Strike48 supports search-in-place connectors for S3, Splunk, and Elastic, enabling direct log reads without data migration. It also offers smart collection for centralized log normalization and AI-assisted parsing that handles semi-structured logs and auto-generates parsers for new sources. **Agent Architecture:** Agents operate using GraphRAG for knowledge grounding, MCP for fine-grained tool exposure, Agent2Agent (A2A) protocols for multi-agent task delegation, and a Workflow Orchestrator for deterministic, auditable sequencing of SOC triage, approvals, and handoffs. **Trigger Mechanisms:** Agents support prompt-based, scheduled, and external event triggers (alerts, vulnerabilities, webhooks). **Infrastructure & Security:** The platform runs on multi-account AWS with Kubernetes, Talos Linux, GitOps, and External Secrets. Edge security is provided via Cloudflare WAF/DDoS/TLS. Runtime isolation uses Cilium network policies, gVisor, and optional WASM sandboxes. LLM access is managed through AWS Bedrock with PostgreSQL and RAG stores. **Deployment Options:** Shared SaaS, isolated compute, and on-premises/air-gapped deployments are available.