CoSupport AI
What is Relari?
Relari provides a contract-based development toolkit for building and deploying reliable AI agents. It allows users to define agent behavior using natural language contracts, enabling systematic verification and iteration.
The platform facilitates rigorous testing through agent simulation and synthetic test case generation, along with providing comprehensive trace analytics for rapid issue identification and resolution. Relari is designed to help teams move quickly and confidently in productionizing AI agents.
Features
- Agent Contracts: Define agent behavior and reasoning in natural language.
- Custom Evaluators: Utilize tailored evaluation metrics to understand agent performance.
- Agent Simulation: Test agent behavior across diverse scenarios.
- Synthetic Test Cases: Generate test cases to enhance testing coverage.
- Trace Analytics: Gain insights into agent execution for rapid issue resolution.
Use Cases
- Develop production grade LLM products.
- Stress test enterprise search engine.
- Quick evaluation of changes in your LLM pipeline.
- Systematically improve AI performance.
- Benchmark and validate agent performance.
FAQs
-
What exactly are Agent Contracts?
They're natural-language specifications that define what an AI agent should do and its reasoning process—like 'check the order policy in the knowledge base and verify customer tier before approving any refund.' Relari turns these specifications into automated tests that verify if your agent follows the intended behavior, helping you catch and fix issues quickly. -
Who should use Agent Contracts?
Teams building and deploying AI agents who need to ensure reliable behavior across scenarios. Natural language allows everyone who understands the user requirements and/or the technical implementations to shape the agents' expected behavior—from product and engineering to QA and domain experts. -
How do Agent Simulation and Synthetic Test Cases work?
Agents need to handle diverse scenarios and edge cases, but manually creating and running all these tests is impractical. Relari lets you create custom simulators for your agents and generate synthetic test cases, multiplying your test coverage while saving time. For example, you can create different customer personas to test your support agent across a wide range of customer intent and interaction patterns. -
Which AI platforms and frameworks are compatible?
Agent Contracts are platform-agnostic work with any agent framework—LangGraph, LlamaIndex, CrewAI, AutoGen, or any custom infrastructure. Simply log your agent's execution traces to Relari with a few lines of code and define your contracts in natural language. Relari handles the verification automatically. -
I can’t have data leave my environment. Can I self-host Relari?
Yes, on our enterprise plan, we offer the option for customers to self-host Relari, ensuring your data remains within your environment.
Related Queries
Helpful for people in the following professions
Featured Tools
Join Our Newsletter
Stay updated with the latest AI tools, news, and offers by subscribing to our weekly newsletter.
