Telemetry policies are now an OpenTelemetry proposal
Tero
Talk to an engineerTry live demo
From the creators of Vector.dev

Make teams own the telemetry they send.

Tero finds bad telemetry, shows who owns it, and turns the fix into a policy your teams can ship.

Works with Datadog, Splunk, OpenTelemetry, and the telemetry stack you already run.

Try live demoTalk to an engineer
app.usetero.com/issues
ISS-1247OpenHighopened 11 min ago

Request logger flooding the hot API path

api-gateway/Platform/Cost/hot-path

Last deploy introduced a request-scoped logger that now fires inside the router middleware. On a path serving ~3.4k req/s, this multiplies baseline log volume by ~25x.

Projected cost+$25,032/ yr
Records45M/ day
Data volume11 GB/ day
Resolvegenerated policy / ready to deploy
DropDrop at the sourcedrop-api-gateway-middleware-trace
id: drop-api-gateway-middleware-tracename: Drop api-gateway middleware tracelog:  match:    - resource_attribute: ["service.name"]      exact: api-gateway    - log_field: LOG_FIELD_BODY      contains: "requestContext entered"  keep: none

Atomic, reviewable, reversible. Stops the bleed now; fix the code on your own schedule.

Log eventrouter_middleware_trace
Log volume45M/day / 11 GB
Resource
service.nameapi-gateway
service.namespaceedge
deployment.environmentprod
Log attributes
event.namerouter_middleware_trace
http.route/v2/checkout
http.request.methodPOST
request.idr_88a1c
Event discoveredFirst seen in the telemetry catalog16 min ago
Hot path check firedA scheduled check flagged this event as an anomaly13 min ago
AI investigationTero verified the anomaly, not a false positive12 min ago
Issue openedPromoted to an issue · policy generated11 min ago
Policy deployedDeploy the generated policy to the estatepending
Resolution verifiedTero confirms the checks clearpending
Problem

Your telemetry scaled.
Your controls still run through you.

More AI agents
More code
More features
More deploys
More services
More telemetry
Manual controls
cost dashboardssampling rulesexclusion filterspipelinestiered storage
A 10x bill spike
Your day, derailed
Renewal shock
Chasing engineers
The cost police
Secrets in logs

Platform is stuck in the middle.

Cost dashboards. Sampling rules. Exclusion filters. Collector config. Pipelines. Storage tiers.

Every team can create the problem. Platform still owns the cleanup.

Discovery

Tero finds bad telemetry and the team that owns it.

Tero watches your logs, detects cost and compliance issues, and shows the service, team, impact, and affected events behind each one.

app.usetero.com
Projected cost$919Kacross 19 services with issues
Estimated waste$390K42% of projected cost
Sensitive data issues76 high priority
Tero identified 8 high-priority issuesReview issues ->
8 services · 24 open issues · 7 sensitive data issuessorted by projected cost
ServiceTeamIssuesDaily volumeProjected cost ▼Estimated waste
api-gatewayPlatform43143 GB / day$197K$85K
kafkaPlatform293 GB / day$158K$73K
frontend-proxyPlatform255 GB / day$112K$48K
checkoutPayments53132 GB / day$107K$43K
product-catalogCommerce256 GB / day$106K$45K
searchDiscovery3101 GB / day$89K$35K
authPlatform4143 GB / day$82K$32K
ordersPayments2140 GB / day$67K$29K
19 services with issues · nothing sampledestimated waste $390K / yr · projected cost $919K / yr
Policy Control

Every telemetry issue becomes
a policy.

A policy is the fix: small, reviewable, scoped, auditable, reversible, and vendor-neutral. Not a dashboard note. Not a buried ticket. Not another hand-written exclusion filter. It is open YAML your team can own, ship, and take with you.

Six properties · no exceptions
01
File-backedplain YAML, not hidden product state
02
Atomicone issue, one policy, one owner
03
Scopedtarget a service, event, route, or attribute
04
Reviewableapprove the change before it ships
05
Portableruns through your existing telemetry tools
06
Reversibleroll back without changing app code

Built on an open standard, donated to OpenTelemetry as OTEP-4738 ->

Runtime

Policies run upstream.

Deploy policies into OpenTelemetry Collector, Vector, Tero Edge, or the pipeline you already run. Drop, sample, redact, or route telemetry before it hits storage, indexing, egress, and vendor bills.

Clean signal
host · ip-10-0-12-44 · us-east-1
OTel Collector
Tero policy engineGo
policy sync·hits·drops·impact
Integration guide

Policies sync to the runtime

Approved policies sync to the runtime. Each runtime reports back hits, drops, and impact. You skip the config rollout, the redeploy, and the ticket.

Policy spec ->

10,000 policies, one pass

The router compiles every policy into a single matching pass, built on Hyperscan, so p99 holds constant from 10 policies to 10,000.

Benchmarks ->Deep dive ->
Workflow

Teams review and ship their own policies.

Platform keeps the guardrails. Tero finds the issue, the owning team reviews the policy, the policy ships through your workflow, and Tero verifies the result.

Your teams

run the workflow themselves
@platform@checkout@payments@commerce
Tero control plane

Tero finds an issue and shows the team that owns it. You are no longer the handoff point for every fix.

Your teams

each reviews and ships its own policies
@platform8 services
@checkout4 services
@payments3 servicesowns issue
@commerce6 services
@search5 services
+6 moresame review path

Tero control plane

live

Your infrastructure

Your services

api-gatewayx8
checkoutx4
paymentsx3

+ 14 more services

Data plane

OTel Collectorcollector
Vectordrop-in
Tero Edgezig · 6mb

+ any conformant runtime

Destinations

Datadogissue found
Splunk
CloudWatch

+ any OpenTelemetry sink

Resolution

Policies fit the way your teams already work.

Use a policy to stop a spike, redact sensitive data, sample noisy events, or route an issue to the owning team. Some policies are the fix. Some give the team room to fix the code.

ISS-1247HOT PATH · COSThigh

Debug logging exploded in the checkout hot path

A debug log in the request path is shipping on every checkout call.

Projected cost
+$25K/yr
Records
45M/day
Owner
@platform
ResponseBlock now, fix later
  1. Rate-limit policy
  2. Alert team
  3. Open Jira
  4. Hand to Claude Code
Outcome

Finally, telemetry that works for everyone.

Leadership gets fewer surprises. Platform gets out of the cleanup loop. Engineering teams own the data they create.

For your leadership

Spend and risk stop becoming surprises

Fewer renewal shocks to explain

A telemetry budget you can forecast

Less sensitive data landing downstream

Spending discipline that holds quarter over quarter

For platform

Get out of the cleanup loop

No more 2am fire drills from a runaway log

Stop chasing teams to clean up after themselves

One place to see every policy: where it runs, what it saves

Keep control without owning every fix

For engineering teams

Own the telemetry your services create

See the context behind every requested fix

An honest mistake stops being an incident

Debug faster. Signal is not buried in junk

Fewer random cleanup pings derailing your sprint

Built by the team behind Vector.dev · OpenTelemetry maintainers · telemetry policies are now an OpenTelemetry proposal