![How to Build an AI Assistant with Keith Moehring [MAICON 2025 Speaker Series]](https://www.marketingaiinstitute.com/hubfs/MAICON-Speaker_Series%20%281%29.png)
![[The AI Show Episode 156]: AI Answers - Data Privacy, AI Roadmaps, Regulated Industries, Selling AI to the C-Suite & Change Management](https://www.marketingaiinstitute.com/hubfs/ep%20156%20cover.png)
![[The AI Show Episode 155]: The New Jobs AI Will Create, Amazon CEO: AI Will Cut Jobs, Your Brain on ChatGPT, Possible OpenAI-Microsoft Breakup & Veo 3 IP Issues](https://www.marketingaiinstitute.com/hubfs/ep%20155%20cover.png)
.jpg?width=1920&height=1920&fit=bounds&quality=70&format=jpg&auto=webp#)
_Michael_Burrell_Alamy.jpg?width=1280&auto=webp&quality=80&disable=upscale#)
Build a Powerful Multi-Tool AI Agent Using Nebius with Llama 3 and Real-Time Reasoning Tools
In this tutorial, we introduce an advanced AI agent built using Nebius’ robust ecosystem, particularly the ChatNebius, NebiusEmbeddings, and NebiusRetriever components. The agent utilizes the Llama-3.3-70B-Instruct-fast model to generate high-quality responses, incorporating external functionalities such as Wikipedia search, contextual document retrieval, and safe mathematical computation. By combining structured prompt design with LangChain’s modular framework, this […] The post Build a Powerful Multi-Tool AI Agent Using Nebius with Llama 3 and Real-Time Reasoning Tools appeared first on MarkTechPost.
In this tutorial, we introduce an advanced AI agent built using Nebius’ robust ecosystem, particularly the ChatNebius, NebiusEmbeddings, and NebiusRetriever components. The agent utilizes the Llama-3.3-70B-Instruct-fast model to generate high-quality responses, incorporating external functionalities such as Wikipedia search, contextual document retrieval, and safe mathematical computation. By combining structured prompt design with LangChain’s modular framework, this tutorial demonstrates how to build a multi-functional, reasoning-capable AI assistant that is both interactive and extensible. Whether for scientific queries, technological insights, or basic numerical tasks, this agent showcases the potential of Nebius as a platform for building sophisticated AI systems.
!pip install -q langchain-nebius langchain-core langchain-community wikipedia
import os
import getpass
from typing import List, Dict, Any
import wikipedia
from datetime import datetimeWe begin by installing essential libraries, including langchain-nebius, langchain-core, langchain-community, and Wikipedia, which are crucial for building a feature-rich AI assistant. It then imports necessary modules such as os, getpass, datetime, and typing utilities, and initializes the Wikipedia API for external data access.
from langchain_core.documents import Document
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_core.tools import tool
from langchain_nebius import ChatNebius, NebiusEmbeddings, NebiusRetriever
if "NEBIUS_API_KEY" not in os.environ:
os.environ["NEBIUS_API_KEY"] = getpass.getpass("Enter your Nebius API key: ")We are importing core components from LangChain and Nebius to enable document handling, prompt templating, output parsing, and tool integration. It sets up key classes such as ChatNebius for language modeling, NebiusEmbeddings for vector representation, and NebiusRetriever for semantic search. The user’s Nebius API key is securely accessed using getpass to authenticate subsequent API interactions.
class AdvancedNebiusAgent:
"""Advanced AI Agent with retrieval, reasoning, and external tool capabilities"""
def __init__(self):
self.llm = ChatNebius(model="meta-llama/Llama-3.3-70B-Instruct-fast")
self.embeddings = NebiusEmbeddings()
self.knowledge_base = self._create_knowledge_base()
self.retriever = NebiusRetriever(
embeddings=self.embeddings,
docs=self.knowledge_base,
k=3
)
self.agent_prompt = ChatPromptTemplate.from_template("""
You are an advanced AI assistant with access to:
1. A knowledge base about technology and science
2. Wikipedia search capabilities
3. Mathematical calculation tools
4. Current date/time information
Context from knowledge base:
{context}
External tool results:
{tool_results}
Current date: {current_date}
User Query: {query}
Instructions:
- Use the knowledge base context when relevant
- If you need additional information, mention what external sources would help
- Be comprehensive but concise
- Show your reasoning process
- If calculations are needed, break them down step by step
Response:
""")
def _create_knowledge_base(self) -> List[Document]:
"""Create a comprehensive knowledge base"""
return [
Document(
page_content="Artificial Intelligence (AI) is transforming industries through ML, NLP, and computer vision. Key applications include autonomous vehicles, medical diagnosis, and financial trading.",
metadata={"topic": "AI", "category": "technology"}
),
Document(
page_content="Quantum computing uses quantum mechanical phenomena like superposition and entanglement to process information. Companies like IBM, Google, and Microsoft are leading quantum research.",
metadata={"topic": "quantum_computing", "category": "technology"}
),
Document(
page_content="Climate change is caused by greenhouse gas emissions, primarily CO2 from fossil fuels. Renewable energy sources are crucial for mitigation.",
metadata={"topic": "climate", "category": "environment"}
),
Document(
page_content="CRISPR-Cas9 is a revolutionary gene editing technology that allows precise DNA modifications. It has applications in treating genetic diseases and improving crops.",
metadata={"topic": "biotechnology", "category": "science"}
),
Document(
page_content="Blockchain technology enables decentralized, secure transactions without intermediaries. Beyond cryptocurrency, it has applications in supply chain, healthcare, and voting systems.",
metadata={"topic": "blockchain", "category": "technology"}
),
Document(
page_content="Space exploration has advanced with reusable rockets, Mars rovers, and commercial space travel. SpaceX, Blue Origin, and NASA are pioneering new missions.",
metadata={"topic": "space", "category": "science"}
),
Document(
page_content="Renewable energy costs have dropped dramatically. Solar & wind power are now cheaper than fossil fuels in many regions, driving global energy transition.",
metadata={"topic": "renewable_energy", "category": "environment"}
),
Document(
page_content="5G networks provide ultra-fast internet speeds and low latency, enabling IoT devices, autonomous vehicles, and augmented reality applications.",
metadata={"topic": "5G", "category": "technology"}
)
]
@tool
def wikipedia_search(query: str) -> str:
"""Search Wikipedia for additional information"""
try:
search_results = wikipedia.search(query, results=3)
if not search_results:
return f"No Wikipedia results found for '{query}'"
page = wikipedia.page(search_results[0])
summary = wikipedia.summary(search_results[0], sentences=3)
return f"Wikipedia: {page.title}n{summary}nURL: {page.url}"
except Exception as e:
return f"Wikipedia search error: {str(e)}"
@tool
def calculate(expression: str) -> str:
"""Perform mathematical calculations safely"""
try:
allowed_chars = set('0123456789+-*/.() ')
if not all(c in allowed_chars for c in expression):
return "Error: Only basic mathematical operations allowed"
result = eval(expression)
return f"Calculation: {expression} = {result}"
except Exception as e:
return f"Calculation error: {str(e)}"
def _format_docs(self, docs: List[Document]) -> str:
"""Format retrieved documents for context"""
if not docs:
return "No relevant documents found in knowledge base."
formatted = []
for i, doc in enumerate(docs, 1):
formatted.append(f"{i}. {doc.page_content}")
return "n".join(formatted)
def _get_current_date(self) -> str:
"""Get current date and time"""
return datetime.now().strftime("%Y-%m-%d %H:%M:%S")
def process_query(self, query: str, use_wikipedia: bool = False,
calculate_expr: str = None) -> str:
"""Process a user query with optional external tools"""
relevant_docs = self.retriever.invoke(query)
context = self._format_docs(relevant_docs)
tool_results = []
if use_wikipedia:
wiki_keywords = self._extract_keywords(query)
if wiki_keywords:
wiki_result = self.wikipedia_search(wiki_keywords)
tool_results.append(f"Wikipedia Search: {wiki_result}")
if calculate_expr:
calc_result = self.calculate(calculate_expr)
tool_results.append(f"Calculation: {calc_result}")
tool_results_str = "n".join(tool_results) if tool_results else "No external tools used"
chain = (
{
"context": lambda x: context,
"tool_results": lambda x: tool_results_str,
"current_date": lambda x: self._get_current_date(),
"query": RunnablePassthrough()
}
| self.agent_prompt
| self.llm
| StrOutputParser()
)
return chain.invoke(query)
def _extract_keywords(self, query: str) -> str:
"""Extract key terms for Wikipedia search"""
important_words = []
stop_words = {'what', 'how', 'why', 'when', 'where', 'is', 'are', 'the', 'a', 'an'}
words = query.lower().split()
for word in words:
if word not in stop_words and len(word) > 3:
important_words.append(word)
return ' '.join(important_words[:3])
def interactive_session(self):
"""Run an interactive session with the agent"""
print("
Read More
