Build a DeepSeek AI Assistant with PAI‑RAG: Internet Search & Enterprise Knowledge Base

Overview

DeepSeek series models have gained global attention for their strong performance, often matching or surpassing top‑tier closed‑source models. Since February 2025, Alibaba Cloud AI platform PAI has released best‑practice solutions for DeepSeek, covering rapid deployment, application building, distillation, and fine‑tuning, enabling developers to efficiently run DeepSeek‑R1, DeepSeek‑V3 and other models on the cloud.

This article presents a solution for building a "DeepSeek + Internet Search + Enterprise Knowledge‑Base Assistant" using PAI‑RAG, which supports one‑click deployment to enterprise WeChat, public WeChat, DingTalk group‑chat bots, and other scenarios to boost business efficiency and user experience.

Step‑by‑Step Guide

Step 1: Deploy a RAG Dialogue System via PAI‑EAS

Open the EAS console, click “Deploy Service”, and select “Deploy Large Model RAG Dialogue System”.

PAI‑RAG offers two deployment modes: an integrated LLM‑RAG deployment (simpler) and a separated LLM deployment (more flexible).

This example uses DeepSeek‑R1‑Distill‑Qwen‑32B to achieve an integrated LLM+RAG deployment.

Step 2: Enable the Internet Search Service

Visit Alibaba Cloud Search Service purchase page, select performance specifications, buy the service, and record the AccessKey and AccessSecret.

Step 3: Configure Network Search in the RAG Application

After deploying the RAG service, open the WebUI and enable the network‑search feature.

Select “Alibaba Cloud” as the search engine, fill in the number of results, service address, AccessKey and AccessSecret.

Choose the chat mode “Chat (Web Search)”.

Step 4: Real‑Time Web Search Conversation Test

Display of web‑search‑enabled responses.

Comparison with responses when network search is disabled.

Enterprise‑Level Capabilities

Private Knowledge‑Base Q&A : Upload proprietary knowledge and let DeepSeek understand internal data via LLM+RAG, supporting multimodal, OCR, embedding, multi‑path retrieval, and re‑ranking.

Multimodal Retrieval & Dialogue : Intelligent parsing of multimodal documents and generation of image‑text replies.

Agentic RAG : Built‑in tools, custom API tools (e.g., inventory, order queries), custom code tools, and multi‑intent routing (retrieval, tool call, DB query, web search).

Database Q&A : Connect existing databases, automatically generate queries, column selection, value retrieval, history lookup, and SQL error correction.

OpenAI Compatibility : Standard OpenAI‑compatible API for seamless integration with OpenAI‑enabled clients such as OpenWebUI, Chatbox, AnythingLLM, Cherry Studio.

Ecosystem Integration : One‑click deployment to enterprise WeChat, public WeChat, and DingTalk bots.

Content Safety Filtering : Built‑in safety filters for text, images, and other risk content.

Elastic Scaling & Multi‑Instance Deployment : Automatic horizontal scaling to handle load spikes.

Tracing : Automatic trace generation for each service call, enabling performance evaluation.

Related Products

PAI‑RAG is an out‑of‑the‑box modular RAG framework that combines LLM reasoning with enterprise‑grade capabilities such as document parsing, splitting, embedding, multi‑turn dialogue, query rewriting, function calling, automated evaluation, content safety, and tracing. It runs on Kubernetes, offering fault tolerance, elastic scaling, and request queuing.

Open‑source repository: https://github.com/aigc-apps/PAI-RAG

PAI Model Gallery aggregates high‑quality pre‑trained models from global AI communities, providing BladeLLM, SGLang, and vLLM acceleration for one‑click deployment of DeepSeek‑V3 and DeepSeek‑R1 series.

Alibaba Cloud General Search Service delivers real‑time open‑domain search, enabling LLMs to fetch up‑to‑date information efficiently.

Best‑Practice Resources

Large Model RAG Dialogue System: https://x.sm.cn/AsCme6o

RAG with Internet Search for AI Q&A: https://x.sm.cn/1OX8MjV

EAS & Elasticsearch RAG: https://x.sm.cn/CNEJO9C

EAS & Milvus RAG: https://x.sm.cn/B2FeNbJ

EAS & RDS PostgreSQL RAG: https://x.sm.cn/7EtIEQU

EAS & OpenSearch RAG: https://x.sm.cn/BEf6iuq

AppFlow + DingTalk AI Bot Guide: https://x.sm.cn/RfOnU2

AppFlow + WeChat Smart Customer Service: https://x.sm.cn/dWHxiI

AppFlow + Enterprise WeChat AI Assistant: https://x.sm.cn/1ZSjKnw

Contact