Skip to main content

Open Source

TypeScript AI AgentFramework

Escape no-code limits and scratch-built overhead. Build, customize, and orchestrate AI agents with full control, speed, and a great DevEx.

Get Started
$npm create voltagent-app@latest
Prompts
API Calls
Webhooks
Context
VoltAgent
AI Resp.
Automations
API Resp.
Webhooks
Workflow
LLM
Memory
RAG
Tools
import { VoltAgent, Agent, createTool, createHooks } from "@voltagent/core";
import { VercelAIProvider } from "@voltagent/vercel-ai";
import { openai } from "@ai-sdk/openai";
import { fetchRepoContributorsTool } from "./tools";
import { fetchRepoStarsTool } from "./tools";

// Create the stars fetcher agent
const starsFetcherAgent = new Agent({
  name: "Stars Fetcher",
  description: "Fetches the number of stars for a GitHub repository using the GitHub API",
  llm: new VercelAIProvider(),
  model: openai("gpt-4o-mini"),
  tools: [fetchRepoStarsTool],
});

// Create the contributors fetcher agent
const contributorsFetcherAgent = new Agent({
  name: "Contributors Fetcher",
  description: "Fetches the list of contributors for a GitHub repository using the GitHub API",
  llm: new VercelAIProvider(),
  model: openai("gpt-4o-mini"),
  tools: [fetchRepoContributorsTool],
});

// Create the analyzer agent
const analyzerAgent = new Agent({
  name: "Repo Analyzer",
  description: "Analyzes repository statistics and provides insights",
  llm: new VercelAIProvider(),
  model: openai("gpt-4o-mini"),
});

// Create the supervisor agent that coordinates all the sub-agents
const supervisorAgent = new Agent({
  name: "Supervisor",
  description: `You are a GitHub repository analyzer. When given a GitHub repository URL or owner/repo format, you will:
1. Use the StarsFetcher agent to get the repository's star count
2. Use the ContributorsFetcher agent to get the repository's contributors
3. Use the RepoAnalyzer agent to analyze this data and provide insights

Example input: https://github.com/voltagent/voltagent or voltagent/voltagent
`,
  llm: new VercelAIProvider(),
  model: openai("gpt-4o-mini"),
  subAgents: [starsFetcherAgent, contributorsFetcherAgent, analyzerAgent],
});

// Initialize the VoltAgent with the agent hierarchy
new VoltAgent({
  agents: {
    supervisorAgent,
  },
});

Read-only

Enterprise-level AI agents

Complete toolkit for enterprise level AI agents

Design production-ready agents with unified APIs, tools, and memory.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
import { Agent } from '@voltagent/core'
import { VercelAIProvider } from '@voltagent/vercel-ai'

import { openai } from '@ai-sdk/openai'
import { anthropic } from '@ai-sdk/anthropic'


const agent = new Agent({
provider: new VercelAIProvider(),
model: openai("gpt-4o-mini"),
});


const anthropicAgent = new Agent({
provider: new VercelAIProvider(),
model: anthropic('claude-3-haiku-20240307'),
});
Tool calling

Enable agents to invoke functions, interact with systems, and perform actions.

Unified API

Seamlessly switch between different AI providers with a simple code update.

Dynamic Prompting

Experiment, fine-tune, and iterate your AI prompts in an integrated environment.

Persistent Memory

Store and recall interactions to enhance your agents intelligence and context.

Intelligent Coordination

Supervisor agent orchestration

Build powerful multi-agent systems with a central Supervisor Agent that coordinates specialized agents.

User
Lead Agent
Supervisor
gpt-4o-mini
Team
AgentA
claude-3.7
AgentB
gpt-4
AgentC
Custom LLM
Conversation-History
User-Lead Memory
Lead-Team Memory
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
import { Agent } from "@voltagent/core";
import { VercelAIProvider } from "@voltagent/vercel-ai";
import { openai } from "@ai-sdk/openai";

// Define supervisor agent
const supervisorAgent = new Agent({
name: "Supervisor Agent",
description: "You manage a workflow between specialized agents.",
llm: new VercelAIProvider(),
model: openai("gpt-4o-mini"),
subAgents: [storyAgent, translatorAgent]
});

// Define story agent
const storyAgent = new Agent({
name: "Story Agent",
description: "You are a creative story writer.",
llm: new VercelAIProvider(),
model: openai("gpt-4o-mini"),
});

// Define translator agent
const translatorAgent = new Agent({
name: "Translator Agent",
description: "Translate English text to German",
llm: new VercelAIProvider(),
model: openai("gpt-4o-mini"),
});

// Stream response from supervisor agent
const result = await supervisorAgent.streamText(
"Write a 100 word story in English."
);

for await (const chunk of result.textStream) {
console.log(chunk);
}
Centralized Coordination
Supervisor Agent manages the workflow, delegates tasks to specialized agents, and maintains context across the entire process.
Specialized Agent Roles
Each agent in the workflow can be optimized for specific tasks, with custom tools, knowledge, and capabilities.
Shared Memory System
Maintain context and state across multiple agent interactions, enabling complex reasoning and multi-step problem solving.
Dynamic Agent Selection
Supervisor intelligently routes tasks to the most appropriate agents based on the current context and requirements.

RAG

Accurate and context-aware responses

For advanced querying and dynamic analysis, integrate data into a knowledge base by syncing from diverse sources

.embed()
.query()
.rerank()
input
embeddingmodel
retrieval
llm
output
Voyage
Voyage
Cohere
Pinecone
Pinecone
Custom
knowledgebase
embeddingmodel
vector stores
OpenAI
Cohere
Voyage
Voyage
Pinecone
Pinecone
PostgreSQL Logo
Postgres
Supabase
Integrated Vector Database
Seamlessly manage and query your vector data with a unified API, compatible with various providers.
Precise Data Filtering
Refine search results by filtering vectors based on metadata like source, date, or custom attributes.
AI Agent Integration
Empower AI agents to access and utilize your knowledge base through dedicated vector search tools.
Hybrid Search
Combine keyword and vector search techniques for enhanced accuracy and relevance in information retrieval.

INTEGRATIONS

Easily connect with 40+ apps in no time

Integrate your AI agents with your preferred tools and services effortlessly.

Lightning Bolt
OpenAI
Anthropic
Anthropic
Notion
Notion
Supabase
Stripe Logo
Stripe
Slack
Slack
Dropbox
Dropbox
Gmail
Gmail
OneDrive
OneDrive
Google Sheets
Google Sheets
Google Drive
Google Drive
Google Calendar
Google Calendar
Microsoft Teams
Microsoft Teams
Mailchimp
Mailchimp
Salesforce
Salesforce
SendGrid
SendGrid
OpenAI
Anthropic
Anthropic
Notion
Notion
Supabase
Stripe Logo
Stripe
Slack
Slack
Dropbox
Dropbox
Gmail
Gmail
OneDrive
OneDrive
Google Sheets
Google Sheets
Google Drive
Google Drive
Google Calendar
Google Calendar
Microsoft Teams
Microsoft Teams
Mailchimp
Mailchimp
Salesforce
Salesforce
SendGrid
SendGrid
OpenAI
Anthropic
Anthropic
Notion
Notion
Supabase
Stripe Logo
Stripe
Slack
Slack
Dropbox
Dropbox
Gmail
Gmail
OneDrive
OneDrive
Google Sheets
Google Sheets
Google Drive
Google Drive
Google Calendar
Google Calendar
Microsoft Teams
Microsoft Teams
Mailchimp
Mailchimp
Salesforce
Salesforce
SendGrid
SendGrid
GitHub
YouTube Logo
YouTube
Zendesk Logo
Zendesk
Trello
Trello
Jira
Jira
Intercom
Intercom
HubSpot
Hubspot
Airtable
Airtable
Figma
Figma
Asana
Asana
Ahref
Ahref
Mixpanel
Mixpanel
Sentry
Sentry
Snowflake
Snowflake
GitHub
YouTube Logo
YouTube
Zendesk Logo
Zendesk
Trello
Trello
Jira
Jira
Intercom
Intercom
HubSpot
Hubspot
Airtable
Airtable
Figma
Figma
Asana
Asana
Ahref
Ahref
Mixpanel
Mixpanel
Sentry
Sentry
Snowflake
Snowflake
GitHub
YouTube Logo
YouTube
Zendesk Logo
Zendesk
Trello
Trello
Jira
Jira
Intercom
Intercom
HubSpot
Hubspot
Airtable
Airtable
Figma
Figma
Asana
Asana
Ahref
Ahref
Mixpanel
Mixpanel
Sentry
Sentry
Snowflake
Snowflake
Observability

Stay in control at every stage

From tracking deployments to debugging and live interaction, VoltAgent gives you full visibility into your AI agents.

Deployment

Deploy your Agents in seconds with VoltAgent Deployment.

Deployment Stages
Initializing build environment
Cloning git repository
Building application
Deploying to VoltAgent's global network
Deployment Logs
10:30:01
Cloning repository...
10:30:02
From https://github.com/your-org/your-app
10:30:02
* branch main -> FETCH_HEAD
10:30:02
HEAD is now at a1b2c3d feat: add new feature
10:30:02
Success: Finished cloning repository files
10:30:03
Using Node.js v18.18.0
10:30:03
Running `npm install`...
10:30:28
Installed 1450 packages in 25.3s
10:30:29
Running `npm run build`...
10:30:29
10:30:29
> tsc && node scripts/postbuild.js
10:30:35
Compiling TypeScript files...
10:30:36
[VoltAgent] Validating agent schemas...
10:30:38
Build successful! Artifacts generated.
10:30:38
[VoltAgent] Starting deployment process...
10:30:40
Uploading build artifacts (2.1MB)...
10:30:45
Upload complete. Verifying deployment...
10:30:46
[VoltAgent] Deployed! 'my-agent-v1.2' is live.
Debugging

Debug and analyze your VoltAgent-powered AI agent's behavior with visual flows.

Observability

Connect your VoltAgent-powered AI agents to popular observability platforms.

LangSmith
Langfuse
Honeyhive Logo
TraceLoop Logo
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
import { VoltAgent } from "@voltagent/core"
import { LangfuseExporter } from "langfuse-vercel"

export const volt = new VoltAgent({
telemetry: {
serviceName: "ai",
enabled: true,
export: {
type: "custom",
exporter: new LangfuseExporter({
publicKey: process.env.LANGFUSE_PUBLIC_KEY,
secretKey: process.env.LANGFUSE_SECRET_KEY,
baseUrl: process.env.LANGFUSE_BASEURL,
}),
},
},
});
AI Chat

Interact with your AI agent through natural language chat interface.

Model
gpt-4-mini
Tools
Social Analysis, Trend Detection, Content Calendar, Performance Metrics, Competitor Analysis
Instructions
You are a marketing assistant specialized in campaign analysis, content strategy, and social media optimization. Analyze data to provide actionable insights and automate routine marketing tasks.

Agent, analyze last month's social media campaign performance and provide recommendations for improvement.

Analysis complete. Last month's campaign reached 45% more users but had 12% lower conversion rate compared to previous campaigns. Key findings: 1) Video content performed 3x better than static images. 2) Posts published between 6-8pm had highest engagement. 3) Product demonstration posts generated most conversions. Recommendation: Increase video content by 40%, focus on product demonstrations, and schedule more posts during evening hours.

Create a content calendar for next month based on these insights. Include optimal posting times and content types.

Content calendar created. I've scheduled 15 posts across platforms with 60% video content focused on product demos. Primary posting times are Tuesdays and Thursdays 6-8pm, with additional posts Monday and Friday mornings based on secondary engagement peaks. I've included hooks for upcoming product launch and integrated seasonal marketing themes. Calendar has been added to your Marketing Projects workspace and synced with team collaboration tools.