Observability

One telemetry layer across every product on the engine.

Events, metrics, and logs from every workflow, plugin, and service flow through a single ingestion layer you control. Self-hosted by default. No third-party SaaS in the data path unless you put one there.

Why centralized observability

Unified Ingestion

One endpoint for all products. Events, metrics, logs — everything flows through the platform.

Automatic LLM tracking

Every LLM call through the Gateway is tracked automatically: tokens, cost, latency, errors.

Product Telemetry SDK

Lightweight client SDK (zero dependencies) for any Node.js or browser product. Batching, retry, flush on exit.

External Source Aggregation

Pull data from cloud billing, CI/CD, external APIs via scheduled workflows. One pane of glass.

Pluggable Storage

SQLite for dev, DuckDB for analytics, ClickHouse for scale. Switch backend in one config line.

Custom Dashboards via Plugins

Platform provides raw data. Your teams build custom visualizations as plugins — no vendor lock-in.

Who benefits

CTO / VP Engineering

Total AI spend, adoption rate, provider efficiency, cost trends

Engineering Manager

Workflow health, fallback rates, team-level usage, error trends

Developer

Request traces, latency breakdown, budget consumption, error details

import { KBTelemetry } from '@kb-labs/telemetry-client';

const telemetry = new KBTelemetry({
  endpoint: 'http://gateway:4000',
  apiKey: process.env.KB_API_KEY,
  source: 'my-product',
});

telemetry.event('user.signup', { plan: 'pro' });
telemetry.metric('api_latency_ms', 142);
telemetry.log('info', 'Payment processed', { amount: 99 });

Stop flying blind across your products

Centralized telemetry. Unified analytics. Your infrastructure, your data.