Currai

Description

Currai is product analytics for teams building conversational AI agents. It monitors chat and voice agents in production, turning real conversations into user stories, intent trends, violations, errors, and evidence-backed improvements, so teams can learn from high-volume conversations without manually reading every transcript.

Its core focus is finding silent failures that users never report. Currai watches every chat and call to catch hidden errors automatically, spot repeated failure patterns, and surface issues as they emerge, such as an agent confirming a refund before the cancellation tool actually completed. It discovers intents and sentiment automatically, tracks intent trends over time, and identifies the moments that drive churn. It also detects quality, policy, and compliance violations, and traces every violation and failed outcome back to the specific conversation with the supporting evidence attached.

Currai sends important agent failures to Slack with evidence already attached, so teams can route alerts by agent or severity, review the evidence without leaving Slack, and open the trace to assign a fix. Getting a first trace takes about five minutes using a coding-agent skill or two native HTTP capture endpoints, with no runtime package or collector to install. It is wire-compatible with Langfuse clients and OpenTelemetry, and Currai hosts ingestion, storage, and dashboards for you. Plans include a free Starter (up to 1,000 events/month, 3-day retention), Pro at $29/month (10,000 events, 14-day retention, full traces and Slack alerts), and Business at $399/month (up to 1,000,000 events, 30-day retention, priority support).

Currai's Core Features

• Silent failure detection across chat and voice agents

• Automatic discovery of intents and sentiment

• Intent trend tracking over time

• Quality, policy, and compliance violation detection

• Full traces with evidence attached to every finding

• Slack alerts routed by agent or severity

• Quick setup via HTTP capture endpoints or a coding-agent skill

• Langfuse- and OpenTelemetry-compatible ingestion

How to use Currai?

1. 1

   Connect capture: Add the Currai coding-agent skill or call the two native HTTP capture endpoints.

2. 2

   See your first trace: The first captured session appears in about five minutes.

3. 3

   Review conversations: Let Currai turn conversations into intents, violations, and errors.

4. 4

   Investigate failures: Trace each violation back to the conversation with attached evidence.

5. 5

   Get alerted: Receive Slack alerts for silent failures and open traces to assign fixes.

Currai's Use Cases

• Silent failure detection
• Intent analytics
• Policy compliance
• Failure investigation
• Team alerting

FAQ from Currai

• What is Currai?

  Currai is product analytics for teams building conversational agents. It shows where users struggle, what they ask for, which product rules break, and the conversations behind every signal.

• What can Currai monitor in production?

  It monitors user stories, intents, violations, errors, tool calls, latency, and cost across production conversations, with trace evidence attached for investigation.

• How long does setup take?

  About five minutes. You add the Currai coding-agent skill or call two native HTTP capture endpoints, with no runtime package or collector to install, and the first session appears immediately.

• Can Currai use my existing instrumentation?

  Yes. Currai accepts native HTTP capture, Langfuse-compatible ingestion, and OpenTelemetry traces, so teams can keep existing instrumentation while adding conversational product analytics.

• Do I have to run any infrastructure?

  No. Currai hosts ingestion, columnar storage, and dashboards for you, so there is nothing to babysit, and it scales with your traffic.

• How is pricing structured?

  Each plan includes a monthly event allowance: Free includes 1,000 events with 3-day retention, Pro is $29/month with 10,000 events and 14-day retention, and Business is $399/month with up to 1,000,000 events and 30-day retention.