# Product recipe: AI assistant over your documents (RAG) (AgentGid)

For an assistant that answers questions from your files (PDF, Office docs, pages) and the web, and cites its sources.

Parts were picked from AgentGid's catalog by rules: free tier, official MCP server, use in public projects built with AI agents, downloads, price. Prices are list prices from the official pages (see each card for the check date); confirm before you commit.

## What it costs
- Start: $0 a month on the free tiers of the first picks (their limits apply).
- If you outgrow the free tiers, the lowest paid plans of the first picks are LlamaCloud (LlamaParse) $50, LangSmith $39 (together about $89/month); usage beyond a plan's credits is billed on top.
- Billed by usage (grows with traffic): Elasticsearch (Elastic Cloud), Exa.
- Model, default: fast and cheap: about $0.85 per 1,000 questions with Claude Haiku 5.5 (AgentGid estimate at ~6,000 input + 500 output tokens each).
- Model, harder questions: about $6.38 per 1,000 questions with Gemini 3.8 Flash (AgentGid estimate at ~6,000 input + 500 output tokens each).
- Prices of the picks below were checked on the official pages between 2026-10-10 and 2026-10-20.

## How it fits together
1. Parse each file into text and chunks; keep the source and page for citations.
2. Embed the chunks and store them in the vector database with their metadata.
3. For a question, search the vector database (and the web search API when the answer may be newer than your files), then send the best chunks to the model.
4. Answer with citations to the chunks used; say so when nothing relevant was found.
5. Trace every call and test with real questions before launch.

## Parse documents (PDF, Office, scans)
- LlamaCloud (LlamaParse): free: 10K credits/month · paid from $50/mo · usage: $1.25 per 1,000 credits · pip llama-cloud · card https://agentgid.com/apis/llamacloud/agent.md
- Azure AI Document Intelligence: free: 0 - 500 pages free per month · usage: Per 1,000 pages: Read $1.50 (0-1M pages), $0.60 (1M+); Layout/prebuilt $10; custom extraction $30; classifier $3 (East US) · npm @azure-rest/ai-document-intelligence · pip azure-ai-documentintelligence · card https://agentgid.com/apis/azure-document-intelligence/agent.md
- Also fits: Adobe PDF Services API, Reducto, Contextual AI, Eden AI (https://agentgid.com/apis/documents/)

## Embeddings
- BGE (FlagEmbedding): open source · pip FlagEmbedding · card https://agentgid.com/apis/bge-embeddings/agent.md
- Voyage AI: free: First 200 million tokens free per account (voyage-4 family, voyage-code-3, voyage-context-4) · usage: $0.02/1M tokens (voyage-4-lite); $0.06/1M (voyage-4); $0.12/1M (voyage-4-large); rerank-3 $0.05/1M · pip voyageai · card https://agentgid.com/apis/voyage-ai/agent.md
- Also fits: OpenAI Embeddings, Mixedbread, Nomic (https://agentgid.com/apis/embeddings/)

## Vector database
- Elasticsearch (Elastic Cloud): open source · usage: Serverless: Ingest as low as $0.14 per VCU per hour; Search as low as $0.09 per VCU per hour; Storage as low as $0.047 per GB retained per month · official MCP https://github.com/elastic/mcp-server-elasticsearch · npm @elastic/elasticsearch · pip elasticsearch · card https://agentgid.com/apis/elastic-cloud/agent.md
- Zilliz Cloud: free: 5 GB storage, 2.5M vCUs per month, up to 5 collections · usage: Dedicated: from $63 per million vectors/month (performance-optimized), from $16 (capacity-optimized), from $5 (tiered-storage); Standard Dedicated from $126/GB/month · official MCP https://github.com/zilliztech/mcp-server-milvus · npm @zilliz/milvus2-sdk-node · pip pymilvus · card https://agentgid.com/apis/zilliz-cloud/agent.md
- Qdrant Cloud: free: Free forever single-node cluster: 0.5 vCPU, 1 GB RAM, 4 GB disk · official MCP mcp-server-qdrant · npm @qdrant/js-client-rest · pip qdrant-client · card https://agentgid.com/apis/qdrant-cloud/agent.md
- Also fits: Chroma Cloud, LanceDB, Weaviate Cloud, Azure AI Search (https://agentgid.com/apis/vector-db/)

## Web search
- Exa: free: $10 credits every month plus a $10 onboarding bonus, no payment method · usage: Search from $4 per 1,000 requests (Lite $4, Auto $7, Deep $12); Contents $1 per 1,000 pages; Monitors $15 per 1,000 requests; Agent $0.012–$1.00 per run · official MCP exa-mcp-server · npm exa-js · pip exa-py · card https://agentgid.com/apis/exa/agent.md
- Tavily: free: 1,000 API credits/month, no credit card · paid from $30/mo · usage: $0.008 per credit (pay as you go); monthly plans $0.0075-$0.005 per credit · official MCP tavily-mcp · npm @tavily/core · pip tavily-python · card https://agentgid.com/apis/tavily/agent.md · MCP server card https://agentgid.com/mcp/servers/io-github-tavily-ai-tavily-mcp/agent.md
- SerpApi: free: 250 searches/month · paid from $25/mo · usage: Starter $25/month for 1,000 searches; Developer $75 for 5,000; Production $150 for 15,000; Big Data $275 for 30,000 · official MCP https://mcp.serpapi.com/mcp · npm serpapi · pip serpapi · card https://agentgid.com/apis/serpapi/agent.md
- Also fits: Parallel, Linkup, Perplexity API (Sonar / Search API), Brave Search API (https://agentgid.com/apis/web-search/)

## Traces and evals
- LangSmith: free: Developer plan: 1 seat, up to 5k base traces / mo, then pay-as-you-go · paid from $39/mo · usage: Plus: $39 / seat per month incl. up to 10k base traces / mo; trace overage is pay-as-you-go (rate not listed on pricing page); 1 LSU = $1 (Deployment/Fleet/Sandbox usage) · official MCP https://docs.langchain.com/langsmith/langsmith-mcp-server · npm langsmith · pip langsmith · card https://agentgid.com/apis/langsmith/agent.md
- LangWatch: free: Developer plan: 50k events/month, 14-day data access, 2 users, no credit card · usage: Growth: EUR 29 per core-seat/month incl. 200k events, then EUR 5 per 100k events (prices shown in EUR only) · official MCP @langwatch/mcp-server · npm langwatch · pip langwatch · card https://agentgid.com/apis/langwatch/agent.md
- Also fits: W&B Weave, Ragas, Pydantic Logfire, AgentOps (https://agentgid.com/apis/observability/)

## Model (all of these can call tools; prices per 1M tokens)
- Default: fast and cheap: Claude Haiku 5.5 ($0.1 in / $0.5 out) · https://agentgid.com/models/claude-haiku-5-5/agent.md · or GPT-6 Luna, MiMo V2.6 Flash
- Harder questions: Gemini 3.8 Flash ($0.75 in / $3.75 out) · https://agentgid.com/models/gemini-3-8-flash/agent.md · or Kimi K2.7 Code, DeepSeek V4 Pro
- Hardest work: Kimi K3 ($3 in / $15 out) · https://agentgid.com/models/kimi-k3/agent.md · or Claude Opus 5.5, Claude Fable 5.1

## Libraries
- `pip install llama-index`: LlamaIndex (MIT) · https://agentgid.com/tools/llama-index/agent.md
- `npm install ai`: AI SDK (Apache-2.0) · https://agentgid.com/tools/ai/agent.md
- `pip install langgraph`: LangGraph (MIT) · https://agentgid.com/tools/langgraph/agent.md
- `pip install qdrant-client`: Qdrant Client (Apache-2.0) · https://agentgid.com/tools/qdrant-client/agent.md
- `pip install pypdf`: PyPDF (BSD-3-Clause) · https://agentgid.com/tools/pypdf/agent.md
- `pip install unstructured`: Unstructured (Apache-2.0) · https://agentgid.com/tools/unstructured/agent.md
- `pip install pydantic-ai`: Pydantic AI (MIT) · https://agentgid.com/tools/pydantic-ai/agent.md

## Agent Skills (read a skill before you install it: skills can run scripts)
- `npx skills add google/skills --skill agent-platform-rag-engine-management`
- `npx skills add Orchestra-Research/AI-Research-SKILLs --skill qdrant-vector-search`
- `npx skills add Jeffallan/claude-skills --skill rag-architect`
- `npx skills add foryourhealth111-pixel/Vibe-Skills --skill embedding-strategies`
- `npx skills add timescale/pg-aiguide --skill pgvector-semantic-search`
- `npx skills add agentscope-ai/OpenJudge --skill rag-eval`

## Coding agent to build it
- OpenAI Codex: free tier, from $8/mo · https://agentgid.com/agents/openai-codex/agent.md
- Claude Code: from $20/mo · https://agentgid.com/agents/claude-code/agent.md
- OpenCode: free tier, from $10/mo · https://agentgid.com/agents/opencode/agent.md

## Also on AgentGid
- PDF tool kit: https://agentgid.com/build/kits/pdf-tool/
- MCP server for your own service: https://agentgid.com/build/kits/mcp-server/

## Before you ship
- Keep API keys on the server, never in the client; set spending limits where the provider offers them.
- Read each service's terms for your use (data retention, commercial use, rate limits).
- Log model calls (traces) and test with your own real inputs before launch.

Recipe page: https://agentgid.com/build/recipes/rag-assistant/ · all product recipes: https://agentgid.com/build/recipes/agent.md