AI Engineering

Instrument adds Opik tracing to an existing Python or TypeScript app and verifies a real trace lands. It is used when you need to add observability to an LLM workflow, then confirm the trace in Opik and return the trace link.

New Analyzer guides adding support for a new AI coding agent analyzer in Splitrail. Use it when implementing parsing for a tool like Copilot, Cline, or another agent with its own data files.

Opik provides SDK reference for tracing LLM apps, span types, framework integrations, threads, and prompt versioning. Use it when you need the right API for instrumentation, flushing traces, or linking prompts to traces.

Opik Diagnose surfaces the online Opik traces and diagnostics issues that need attention, ranked by errors, tool-call failures, latency, regressions, and low eval scores. Use it to triage production agent behavior before handing the top trace to Opik Explain.

Opik Evaluate helps build and run LLM evaluations for apps, datasets, judges, RAG metrics, synthetic data, and error analysis. Use it to measure AI product quality and validate evaluators against human labels.

Opik Explain roots cause a specific Opik trace or trace pattern and ties the explanation to evidence spans. Use it to debug failed, wrong, or slow traces without changing code.

Pricing guides updating model cost data in Splitrail. Use it when adding new AI models, aliases, caching rules, or dated pricing overrides.

Splitrail MCP lets AI assistants query usage statistics, model breakdowns, cost data, and file operation counts through an MCP server. Use it when adding or modifying tools and resources for the interface.

Agentic Testing Patterns defines how AI agents plan, generate, execute, and maintain test suites with minimal human intervention. Use it for autonomous test orchestration, failure feedback loops, and risk-based test prioritization.

AgenticMail provides AI agents with email, SMS, phone call control, Telegram, media, memory, storage, and task delegation tools. Use it when an agent needs a persistent inbox, outbound messaging, or coordination with other agents.

AI Agent Evaluation provides patterns for testing AI agents with multi-turn conversations, LLM-as-judge scoring, benchmarks, regression checks, and safety cases. Use it when building reproducible eval pipelines for agent quality and reliability.

List the assumptions behind a current approach, flip each to its plausible opposite, and identify where the flipped version is actually true — uncovering new creative directions. Ideal when requirements or designs feel inevitable or unquestioned.