v2.4 shipping · now with distributed tracing

Developer analytics done right —
stop opening Grafana at 2am.

One lightweight SDK for every language. Logs, metrics, traces, and alerts — correlated automatically. Ten-second installation, five-minute value, zero YAML.

~/atlas-api connected · byteflow.app
$ npx byteflow init
▸ Detecting project... found Node.js v20 + Fastify 4
▸ Installing byteflow@2.4.0...
▸ Instrumenting handlers, queues, and database...
✓ Done in 8.2s · 4,142 call-sites instrumented

$ byteflow deploy --watch
▸ Sending telemetry to project atlas-prod...
✓ Live · 412 events/sec · p95 34ms · 0 errors last 5min
⚠ Anomaly detected: /api/billing latency 2.4× baseline
└─ Open: byteflow.app/trace/abc-123

◐ Atlas API · Production

last 5 min · live
Request / sec
412/s
p95 latency
34ms
Error rate
0.23%
Apdex score
0.96
P95 latency by endpoint · 1hr
/api/billing · 2.4× baseline
// Why teams switch

Every boring observability task, handled.

Byteflow replaces your logs, metrics, traces, and uptime checks with a single cohesive tool — priced honestly, instrumented automatically.

// 01 · One-line SDK

Ten seconds to live production data.

No YAML, no collector, no side-car. A single import sends structured logs, metrics, and traces correlated by request-id out of the box.

// atlas-api / server.ts
import { byteflow } from 'byteflow/node';
byteflow.init({ apiKey: 'bf_live_...' });

// That's it. Your logs, metrics, and traces are now on byteflow.app
// 02 · Anomaly watch

We notice before your customers do.

Per-endpoint baseline detection. When p95 crosses 2× its 14-day average for 90 seconds, an alert lands — with the trace that caused it.

2.4×
  • 42,000 teams
  • 6.4B events / day
  • 28% fewer pages for on-call
// 03 · Distributed traces

Every span, every hop, auto-correlated.

OpenTelemetry-native. Adopt our SDK, keep your existing OTel setup, or run both.

// 04 · Structured logs

Search a billion lines in milliseconds.

Indexed on the first character. Query language that looks like SQL but reads like English.

// 05 · On-call, calmly

Alerts that know when to shut up.

Rate-limited per-incident by default. Paging rules you can actually read. Statuspage built-in.

// 06 · Integrations

Works with what you already ship.

Send from Node, Go, Python, Rust, Ruby, PHP, Elixir, Deno, Bun, or any OTel-compliant emitter. Forward alerts to Slack, PagerDuty, or Opsgenie in one click.

NODE
GO
PYTHON
RUST
RUBY
PHP
ELIXIR
BUN
OTEL
SLACK
PD
+40
// 07 · On us

Honest, flat pricing.

$29 / month per service. No "per-host fee", no "enterprise only" features. Your invoice will not be the thing that pages you.

SOC2 · Type IIGDPR · readyEU data · yes

"We migrated 14 services from three separate tools to Byteflow over a weekend. On-call pages dropped 34% the first month. Our infra bill dropped 62%."

A. Nakamura · Staff SRE, Meridian Analytics · 240 engineers

"Byteflow finds the one line of our Python that blew up a database, before our users tweet about it. That is the whole product, shipped. I don't need the other thirty things."

D. Chen · CTO, Orbit.fm · 18 engineers
// start.byteflow.dev

Stop re-implementing observability.

Ten seconds to live production data. Fourteen days free, no card. Cancel from the CLI if you don't love it.