‹ Cancel and return

New workflow

Connect an existing agent that's already running, or start fresh.

Greenfield: design from scratch

Pick a vertical, write the brief, generate a baseline. ~5 min to first run.

Connect existing agent ← you are here

We watch your agent's traces, infer the workflow, and propose improvements after ~100 traces.

1 Connect source

2 Inferred workflow

3 Generated eval set

Pick a trace source

OpenTelemetry endpoint

Point your existing OTel collector at our HTTPS endpoint. Use the gen-ai semantic conventions. No code changes if you already have OTel.

recommended live stream

Upload trace export

JSONL, OTel-protobuf, or LangSmith / Phoenix export. Good for one-shot analysis without wiring up live ingest.

batch no code

Install our SDK

Python / Node / Go wrapper. One-line install. Adds typed event hints (skill-loaded, monitor-signal) on top of plain OTel.

richest signal code change

Configure OTel endpoint

Workflow name

A short label for what this agent does. You can rename later.

Endpoint URL

Send OTLP/HTTP requests here. We accept gen-ai semantic conventions out of the box.

Workspace API key

Set as x-api-key header on all requests. Rotate key →

PII redaction

Applied client-side at our edge before traces hit storage. Configurable per workflow.

Quick start

# Add to your existing OTel collector config
exporters:
  otlphttp/ownevo:
    endpoint: "https://ingest.ownevo.ai/v1/traces"
    headers:
      x-api-key: "oev_acm_••••••••••••••••8d3f"
service:
  pipelines:
    traces:
      exporters: [otlphttp/ownevo]

Drop this into your existing collector config. We'll start receiving spans within seconds.

Live ingest preview

Connected · 38 traces in last 60s

12:04:21 trc_84f2c1 order_intake.run 14 events

12:04:19 trc_84f2c0 order_intake.run 11 events

12:04:17 trc_84f2bf order_intake.run 17 events

12:04:14 trc_84f2be order_intake.run 12 events

12:04:11 trc_84f2bd order_intake.run 15 events