AI Toolkit

Build multi-turn conversational chatbots with memory, context management, and tool-use hooks. Works with OpenAI, Anthropic, or any OpenAI-compatible API.

Generate, cache, and manage text embeddings from any LLM provider. Batch processing, deduplication, and export to common vector formats included.

Full retrieval-augmented generation pipeline: ingest documents, chunk, embed, retrieve, and generate answers. Swap any vector store or LLM backend.

Template library + testing harness for prompt design. A/B test prompts, score outputs, and track performance across model versions.

Set up and query a local vector store (FAISS/Chroma/Qdrant) in minutes. Includes cosine similarity search, metadata filtering, and result ranking.

Build autonomous AI agents with tool-use, planning loops, and memory. Drop-in ReAct and chain-of-thought patterns. Works with any LLM API.

Evaluate LLM outputs across accuracy, relevance, tone, and safety dimensions. Supports BLEU, ROUGE, G-Eval, and custom scoring functions.

Prepare datasets, upload training files, and manage fine-tuning jobs via OpenAI and compatible APIs. Includes dataset quality checks and cost estimation.

Track token usage and API spend across providers. Set budgets, get alerts when thresholds are hit, and export cost breakdowns by project or team.

Detect AI-generated text using perplexity analysis and burstiness scoring. CLI tool with JSON output, batch mode, and configurable confidence thresholds.