Roadmap
Delivered · v0.1
Open-source AI-native OpenTelemetry APM — the core loop is working end to end.
Next Phase · Core Plans
Five priorities to complete the full collect → observe → intelligently act pipeline:
| # | Direction | Goal |
|---|---|---|
| ① | One-Agent open source | Release a unified collection agent for traces, metrics, logs, and host metrics — ready to connect to DataBuff Ingest out of the box |
| ② | OpenTelemetry logs | Ingest OTLP Logs, support OpenTelemetry log collection and trace correlation, completing the three pillars of observability |
| ③ | eBPF zero-instrumentation APM | Kernel-level collection via eBPF with zero code changes for host, container, and K8s scenarios — covering runtime areas agents cannot reach |
| ④ | Agent observability | Observe AI agents themselves — call chains, token usage, tool calls, latency, and errors — so agent runs are traceable and diagnosable |
| ⑤ | AI agent memory | Persistent memory for in-platform multi-agents — retain context and diagnostic conclusions across sessions so AI learns your system over time |
Near-term Plans
| Direction | Goal |
|---|---|
| Deeper AI | Streaming chat, more digital experts, causal root-cause analysis |
| Richer APM | Log correlation, RUM, broader middleware coverage |
| Stronger alerting | Composite rules, incident workflows, notification integrations |
| More stable deployment | Multi-node clusters, persistence options, Helm Chart |
Long-term Vision
From "AI-assisted viewing" to "AI-autonomous management" — that is DataBuff's ultimate direction.