Not Diamond

Key Features

• Prompt Optimization

  Automatically iterates on static prompt templates across models and providers using a data-driven algorithm.

• Intelligent Routing

  A low-latency runtime tool that predicts which model to use for a specific input to balance accuracy, cost, and latency.

• Agent Optimization

  Uses self-improving algorithms to optimize multi-step workflows and autonomous agent trajectories.

• Continuous Learning Layer

  A framework for AI agents and prompts to improve based on evaluation datasets and reinforcement learning.

Use Cases

• Multi-Model Query Routing

  Using intelligent routing to predict and select a model for each incoming user request.

• Automating Prompt Engineering

  Optimizing static prompt templates across different providers to help improve output accuracy.

• Improving Agent Workflows

  Applying self-improving algorithms to support the success rates of multi-step autonomous agents.

FAQ

How does Not Diamond's intelligent routing work?

It uses a low-latency algorithm to predict which LLM should be used for a specific input based on a mapping of inputs to model rankings, allowing for trade-offs between accuracy, cost, and latency.

What are the costs associated with prompt optimization?

While Not Diamond offers a set number of free successful optimizations per month, the actual LLM inference costs incurred during the optimization process are passed through to the user.

Is Not Diamond suitable for enterprise security standards?

The platform is SOC-2 and ISO 27001 compliant and offers VPC deployments and custom ZDR policies for enterprise teams.