Veridex vs LlamaIndex

LlamaIndex is the best-in-class data framework for LLMs. It excels at building indexes, retrieval pipelines, and Q&A over documents. It is not primarily an agent framework.

Dimension	LlamaIndex	Veridex
Primary purpose	Index + retrieval	Agent runtime
Tools	Yes, but secondary	First-class typed contracts
Policy	Ad-hoc	Engine + packs
Memory	Index-shaped	Tiered + reconciler
Approvals	None	First-class
Audit	Logs	Hash-chained, signed bundles

Compose them

LlamaIndex sits behind a Veridex tool:

import { tool } from '@veridex/agents';
import { z } from 'zod';
// import { VectorStoreIndex, Settings } from 'llamaindex';
 
const search = tool({
  name: 'search_kb',
  safetyClass: 'read',
  description: 'Search the corporate knowledge base.',
  input:  z.object({ q: z.string() }),
  output: z.object({ chunks: z.array(z.string()) }),
  async execute({ input }) {
    const result = await llamaIndexQuery(input.q); // returns chunks
    return { success: true, llmOutput: { chunks: result } };
  },
});

The agent gets retrieval; we keep the governance, sandbox, audit.

Choose

Just retrieval / RAG? LlamaIndex.
Agentic workflow with retrieval as one tool? Veridex + LlamaIndex behind a tool.

vs LangGraph vs Mastra