11.4M
spans / sec / node
Sustained ingestion on commodity hardware. No tiered storage trick, no warm-cache marketing.
Live · 9.3T spans ingested this week
OTLP / gRPC · region us-east-1
One observability platform. OpenTelemetry-native. Built for the engineer who gets paged at 3am.
CoreWatch replaces four fragmented monitoring tools with a single OTel-native platform that cuts MTTR by 58% and saves the average customer $214,000 a year. No proprietary agent. No vendor lock-in. No dashboards nobody reads.
Founder note · 2019–present
Observability should not be a stack of dashboards you swear at during an incident. It should be a single, fast surface that tells you — within seconds — what is broken, why it broke, and what to do next. We built CoreWatch after a decade of being on the other side of the page: tired of paying for four tools that disagreed about latency, tired of agents that ate our CPU, tired of writing runbooks no one would read.
We did not set out to make a pretty dashboard. We set out to make fewer pages, faster resolution, and a bill we could defend in a finance review.
Priya Ramanathan
Co-founder · ex-Google SRE
Marcus Chen
Co-founder · ex-Cloudflare Platform Lead
The Columnar Adaptive Index
The benchmark your platform engineer will ask about, answered first.
Two numbers actually matter when you are staring at a live outage: how fast the system asks questions, and what the bill looks like in a year. Here is both.
58%
MTTR reduction
Mean time to resolution across 600+ production customers, Q4 2024 internal benchmark. Reproducible on request.
$214K
saved per year, on average
vs. legacy APM stacks of equivalent scope — Datadog, New Relic, Dynatrace equivalents side-by-side.
9.3T
spans ingested this week
Across the CoreWatch fleet, week of January 2025. Live counter on the status page if you want to watch it move.
Methodology ·
Internal benchmark, Q4 2024. Customer identifiers available under NDA.
In production today
1,847 engineering teams across 41 countries run on CoreWatch.
A sample of named customers who have signed release forms, plus the YC W23 cohort that grew on the platform.
RipplingHR & IT platform
MercuryBanking for startups
LinearIssue tracking
YC W23 cohort4 of the top 10 fastest-growing
# Point the OTel collector at CoreWatch.
receivers:
otlp:
protocols:
grpc: { endpoint: 0.0.0.0:4317 }
http: { endpoint: 0.0.0.0:4318 }
exporters:
otlp/corewatch:
endpoint: ingest.corewatch.net:4317
auth:
authenticator: bearertokenauth
sending_queue:
num_consumers: 16
service:
pipelines:
traces:
receivers: [otlp]
processors: [batch, memory_limiter]
exporters: [otlp/corewatch]OpenTelemetry-native, day one
Wire it in once. Never write a proprietary agent again.
CoreWatch speaks OTLP out of the box. Point the OpenTelemetry Collector at ingest.corewatch.net:4317 and your traces, metrics, and logs arrive in the same shape they left your service — no translation layer, no re-instrumentation when you leave.
The Observe SDK we published as open source in 2021 ships with 4.2M+ monthly PyPI downloads. The collector config above runs on your laptop in under a minute.
- OTLP/gRPC · traces, metrics, logs in one pipeline
- Tail-based sampling · keep the traces that matter
- No agent tax · under 12 MB RSS on the host
30 minutes · live · with a solutions engineer
Bring your own telemetry. We will show it back to you.
No slideware. No deck. You point an OTLP exporter at a sandbox, we open CoreWatch on the call, and you watch your own spans land and resolve. If it is not faster than what you run today, you keep your stack and we buy the coffee.