Skip to content

Frequently Asked Questions

Everything you need to know about the Anosys Platform — from getting started and integrations to analytics, security, pricing, and support.

General

What is Anosys?

Anosys is a Universal Observability as a Service platform. It collects telemetry — traces, metrics, logs, and custom data fields — from any source, processes it through built-in analytics and ML models, and surfaces actionable insights in real time.

Anosys is purpose-built for modern, data-intensive workflows like agentic AI, but works equally well for infrastructure monitoring, web analytics, IoT, and custom business metrics.

Who is Anosys designed for?

Anosys is built for engineering, ML, data, product, and security teams that need to monitor complex workflows — including AI agents, LLM-based applications, APIs, transactions, and IoT devices — and detect performance issues, anomalies, or cost overruns.

How is Anosys different from tools like Datadog or Arize?

Unlike general-purpose observability platforms, Anosys unifies:

  • Custom data ingestion — HTTP pixels, OpenTelemetry, REST APIs, and native SDKs
  • Advanced analytics — statistical and ML-driven anomaly detection
  • Custom pipelines — real-time and batch data processing
  • Evals & model monitoring — track model quality, drift, and calibration
  • KPI dashboards — custom charts, alerts, and business-level reporting

It's purpose-built for AI telemetry and business logic monitoring, not just infrastructure metrics.

Do I need to be a data scientist or ML Engineer to use Anosys?

No. While the platform offers advanced data science tools, it's built to be accessible for any person that wants to understand their data — engineers, product managers, data analysts, data scientists, executives, etc. — with prebuilt templates, dashboards, and low-code pipelines. If you want to go deeper, the full ML toolkit is available.

Agentic AI Observability

Which AI agent frameworks does Anosys support?

Anosys currently supports:

  • OpenAI Agents — auto-instrument the OpenAI Python and JavaScript SDKs with a few lines of code
  • OpenAI Agents SDK — deeper integration with the OpenAI Agents SDK for multi-agent workflows
  • Claude Code — one-command SDK hook installation for full session observability (150+ fields)
  • Claude Code (OTEL-only) — zero-code observability via OpenTelemetry configuration

Any framework that exports OpenTelemetry data can also send telemetry directly to Anosys. See Custom LLM Integrations.

What data does Anosys capture from AI agents?

Depending on the integration, Anosys captures:

  • Traces — every model invocation, tool call, and reasoning step
  • Token usage — input/output token counts per request and per session
  • Latency — time-to-first-token, per-call latency, and end-to-end session duration
  • Model parameters — temperature, top-p, max tokens, model name and version
  • Errors — API failures, rate limits, retry counts, and finish reasons
  • Cost estimates — per-request and per-session cost attribution

The OpenAI SDK integration captures additional data points like tool call arguments, streaming events, and response metadata. See the OpenAI Agents guide for details.

Do I need to modify my agent code?

It depends on the framework:

  • Claude Code (SDK hook) — No code changes. Install the SDK and run anosys-claude-code install. The hook fires automatically. See the setup guide.
  • Claude Code (OTEL-only) — No code changes. Configure ~/.claude/settings.json and traces flow automatically. See the OTEL guide.
  • OpenAI Agents — Add 2–3 lines of Python or JavaScript to initialize the Anosys logger before creating your OpenAI client. See the setup guide.
Can I monitor the Claude Code Desktop App?

Yes. Configure the Desktop App to run the claude CLI over SSH (to a remote machine or even localhost). If the CLI host has the AnoSys Claude Code SDK installed, all Desktop App sessions are automatically observed. See the SSH integration guide.

Does Claude Code observability support content redaction?

Yes. The AnoSys Claude Code SDK offers a built-in content redaction option. When enabled, all user prompts, assistant responses, and thinking blocks are replaced with REDACTED before being sent to Anosys. Only metadata and token counts are transmitted. You can enable redaction during installation with the --redaction flag. See the Claude Code guide.

Can I compare costs across different models?

Yes. Anosys tracks token usage and model identity per request, so you can compare cost and performance side-by-side across models (e.g. gpt-5 vs gpt-4.1-mini, or claude-sonnet-4 vs claude-haiku).

Data Collection & Integration

What data sources can Anosys connect to?

You can send data to Anosys via:

  • REST API — HTTP GET/POST with custom query parameters
  • OpenTelemetry — OTLP/HTTP and OTLP/gRPC endpoints
  • Python SDK — native logger for OpenAI, plus decorator-based logging for custom functions
  • JavaScript — lightweight tracking scripts for web applications
  • Image pixels — 0×0 tracking beacons for web and mobile
  • Bulk ingestion — BigQuery, AWS, and GCP Pub/Sub connectors

See Data Ingestion Options for full details and code examples.

Can I send custom data fields?

Yes. You can define your own schema or use the default telemetry model (e.g. trace_id, timestamp, latency_ms, model_name, user_id). Custom string, numeric, and boolean fields are supported via query parameters or SDK attributes.

Does Anosys support OpenTelemetry (OTEL)?

Yes. Anosys natively supports OTLP/HTTP and OTLP/gRPC endpoints. You can forward spans, metrics, and logs directly from any existing OTEL setup — including the OpenTelemetry Collector, Jaeger, or framework-native exporters.

How often is data processed?

Data streams are ingested and available in real time. Batch pipelines can also be scheduled at minute, hourly, daily, or custom intervals.

Analytics & Detection

What kind of anomalies can Anosys detect?

Anosys provides statistical and ML-based anomaly detection for:

  • Latency spikes and degradation
  • Token usage outliers and cost anomalies
  • Model drift and quality regression
  • User behavior changes
  • Fraud and abuse patterns

You can use built-in detection models or define your own custom rules.

Can I define my own detection rules?

Yes. You can create custom heuristics and thresholds through the platform UI, or define business-specific rules using the pipeline framework.

Does Anosys support machine learning pipelines?

Yes. You can deploy, monitor, and evaluate ML models using the built-in pipeline framework, or integrate your own model endpoints for scoring and evaluation.

Can I visualize data and metrics?

Yes. The dashboard builder supports time-series charts, histograms, scatter plots, tables, gauges, and more — for any ingested metric or field. All dashboards are fully customizable and support real-time updates.

Security & Compliance

How secure is my data?

Anosys uses AES-256 encryption in transit and at rest, supports SOC 2 compliance, and offers RBAC (Role-Based Access Control) for multi-user environments. API keys are scoped per pixel to limit access.

Where is the data stored?

Your data is stored securely in the cloud region you select — U.S., EU, or APAC. Anosys is powered by Google Cloud for state-of-the-art scalability, reliability, and security.

Do you process personally identifiable information (PII)?

No, unless you explicitly send it. We recommend masking or hashing sensitive fields before ingestion to comply with GDPR, CCPA, or other applicable regulations.

Pricing & Plans

How does Anosys pricing work?

Pricing is based on monthly data volume (GB stored and processed). No per-seat charges — your entire team can use the platform. Plans include:

  • Free — Free forever with basic observability features
  • Hobby — For individual developers and side projects. Includes AI Copilot for data insights.
  • Team — For engineering teams. Adds AI Alerts & Auto-Judge, AI Platform Assistant, AI Agent Root Cause Analysis, Hydrated Calls, unlimited metrics, and priority support.
  • Pro — For larger organizations. Adds ML Model Support, Integrity & Governance Rules, and advanced analytics.
  • Enterprise — Custom pricing with AI Workflow Optimization & Cost Intelligence, custom integrations, dedicated support, and SLA guarantees.

See the pricing page for full details.

Do you offer a free tier?

Yes. All users get access to a free forever plan with basic observability features. Paid plans start at the Hobby tier and unlock AI-powered capabilities like the AI Copilot, Auto-Judge, and Root Cause Analysis.

Can I cancel anytime?

Yes. Subscriptions can be canceled or downgraded anytime from the Anosys Console settings.

Getting Started

How do I start sending data to Anosys?
  1. Sign up at console.anosys.ai
  2. Create a pixel (choose your integration type)
  3. Follow the integration guide for your use case:
Is there an API reference or developer docs?

Yes. Full documentation is available here, including SDK setup, integration guides, code examples, and architecture diagrams.

Do you provide onboarding support?

Yes. Our team offers Slack-based technical onboarding for business accounts, plus detailed tutorials and walkthroughs. We'll make sure the platform is configured to fit your needs.

Integrations & Exports

Does Anosys integrate with alerting tools?

Yes. You can send alerts to Slack or email based on thresholds, anomaly detections, or custom rules.

Can I export raw data?

Yes. You can export data via API or CSV from any dashboard or query result.

Technical Support

How can I reach support?

Email support@anosys.ai or use the in-app chat. Enterprise clients get access to dedicated Slack channels and on-call support.

What's your uptime guarantee?

Anosys provides a 99.9% uptime SLA for production-tier plans.

How often is the platform updated?

New releases are deployed biweekly with security patches, performance improvements, and feature enhancements.

Last updated: March 8, 2026