Changelog

AI Root Cause Analysis — ask “why did this fail?” in natural language. The reasoning engine traces across your entire service graph to pinpoint the exact method and line of code.

Execution Time Machine — replay any distributed request step by step. Scrub forward and backward through time with full data context at each hop.

Natural Language Queries — ask your system anything in plain English. "How does checkout work?" "What depends on InventoryDB?" "What changed since yesterday?"

Live Event Streaming — real-time event feed showing method calls, DB queries, and service interactions as they happen.

Stripe billing integration — usage-based billing with automatic metering. Free tier for teams up to 5 engineers.

Kafka consumer rebalancing — fixed an issue where agent data ingestion experienced delays during pod restarts due to consumer group rebalancing. Added graceful shutdown hooks.

Prometheus + Grafana monitoring — comprehensive dashboards for all services including Kafka lag, Graph Database query latency, reasoning pipeline throughput, and agent collector health.

Graph database connection pooling — resolved memory leak in long-running graph traversals. Connection pool now properly releases idle connections after 30s.

Cypher query optimization — rewrote top 5 most expensive knowledge graph queries. Average query time reduced from 340ms to 48ms for graph traversals with 3+ hops.

Usage analytics service — real-time tracking of investigation count, token consumption, active users, and graph size per tenant. Powers the new billing dashboard.

Python debugger agent rewrite — replaced sys.settrace with a bytecode-level instrumentation approach. Overhead reduced from 12% to under 3% in production benchmarks.

Investigation API v2 — the POST /v1/investigations endpoint now requires a timeRange field. Requests without it will return 400 Bad Request.

Docker build CVE fix — updated base images to resolve critical CVE in libcurl. Removed unnecessary npm registry dependencies from landing page Dockerfile.

Semantic Knowledge Graph — the core of NxtOne. Runtime execution captured as an ontology graph in Graph Database with services, classes, methods, databases, and queues as nodes with typed relationships.

Cross-service tracing — follow any request across every microservice it touches with zero manual instrumentation. Auto-correlation via trace context propagation through HTTP headers and Kafka message metadata.

Java agent (Spring Boot) — production-ready agent with @NxtOneAgent annotation. Bytecode instrumentation via ASM. Less than 3% CPU overhead in benchmarks.

Python agent (FastAPI) — ASGI middleware-based agent. Captures async execution flows with proper coroutine tracking. pip install nxtone-agent.

AWS EKS deployment — full production infrastructure with Helm charts, horizontal pod autoscalers, and Terraform modules for VPC, EKS cluster, and managed node groups.

Event-driven architecture — Kafka-based pipeline for agent data ingestion. Events flow from agents through the collector, into Kafka topics, and are consumed by the graph builder and analytics services.

Multi-database architecture — Graph Database for the knowledge graph, Couchbase for session data, Redis for caching and real-time state. Each database chosen for its strengths.

CI/CD pipeline — Jenkins pipelines with automated testing, Docker builds, vulnerability scanning, and staged deployments to EKS. Feature branch previews with isolated namespaces.

Proof of concept — first working capture of Java method calls as semantic graph relationships. Validated that AI models can reason about execution behavior when given structured ontology data instead of raw logs.

Graph database graph model v1 — initial schema design with :Service, :Method, :Database, and :CALLS, :QUERIES, :EMITS relationship types. The foundation everything else is built on.