Description
AGNT is a local-first agent operating system designed to centralize and manage AI agents, workflows, and goals directly on your own hardware. It provides a unified workspace where users can integrate their existing AI subscriptions, API keys, and local models like Ollama and LM Studio. The platform, featuring an AI assistant named Annie, coordinates diverse tasks across research, code, content, and operations, offering memory, automation, and cost visibility.
Read More
The core functionality revolves around persistent AI workers (agents) with their own memory, assigned tools, credit limits, and model routing. These agents learn from prior runs, improving performance over time. AGNT also includes a visual workflow builder with over 119 tool nodes, supporting triggers, branching, and version rollback. This allows for deterministic execution where precise control is needed and agentic behavior where flexibility is beneficial. Users can create, run, import, and version workflows, with full version history and one-step rollback capabilities.
AGNT emphasizes local-first operation, meaning Community Core runs entirely on your machine, keeping data, prompts, code, credentials, and receipts local unless a remote cloud provider is explicitly connected. API keys and tokens are stored in the operating system's encrypted keychain, injected at execution time, and never written to logs or prompts. For sensitive actions like sending emails or changing records, approval gates can be set up, and every action is recorded in a verifiable receipt.
The platform supports a wide range of integrations, including over 50 built-in options, any MCP server, any REST API, and 6,000+ apps via Zapier. It can delegate work to external engines such as Grok CLI, Codex, Claude Code, Hermes, and OpenCode, allowing users to coordinate broader work around existing coding agents. AGNT also offers cost reduction through caching, which can significantly lower API usage expenses.
AGNT is available as a free local desktop app for personal use, education, nonprofits, and small businesses. Paid plans offer dedicated cloud instances, encrypted vault backup, remote agent execution, team collaboration features, and enhanced services like hosted models, search, scraping, sandboxed environments, email, and webhooks. The platform can be run as a desktop app or a Docker image, with the Community Core source available under a specific license for auditing and modification.
- CategoryAI Agent Builders
- Pricing ModelFreemium
- DeploymentDesktop App, Docker, Self-hosted, Cloud
- PlatformsWindows, macOS, Linux
AGNT's Core Features
Local-first agent operating system
Unified workspace for AI subscriptions, models, and agents
Visual AI workflow builder with 119+ tool nodes
Persistent AI agents with memory, tools, and credit budgets
Support for local models (Ollama, LM Studio) and cloud APIs
Cost visibility and caching for AI cost reduction
Encrypted local storage for API keys and tokens
Approval gates for sensitive agent actions
Verifiable receipts for every agent run
Integration with 50+ built-in apps, REST APIs, and Zapier (6,000+ apps)
Self-hosting options via desktop app or Docker
Version history and rollback for workflows
Delegation to external agent engines (Codex, Claude Code, Grok CLI)
Community Core license for personal and small business use
Dedicated cloud instances with paid plans
How to use AGNT?
Install AGNT: Download and install the free local desktop app for Windows, macOS, or Linux, or run it via Docker.
Connect Models: Configure your preferred AI models by connecting existing API keys from providers like OpenAI or Anthropic, or by pointing to local models like Ollama or LM Studio.
Create Agents: Define persistent AI agents, assigning them specific tools, memory, and credit budgets for various jobs.
Build Workflows: Use the visual workflow builder to design automated processes with triggers, branching logic, and a library of tool nodes.
Grant Authority: Set approval gates for sensitive actions to ensure agents operate within defined boundaries.
Execute Tasks: Give agents jobs to perform, such as research, code generation, content creation, or operational tasks.
Monitor and Review: Inspect detailed receipts for every run, tracking actions, costs, and failures, and review version history for workflows.
Optimize Costs: Leverage built-in caching and subscription economics to reduce overall AI operational expenses.
AGNT's Use Cases
- Inbox Triage
- Support First Responder
- First-Pass Reviewer
- Research Analyst
- Competitive Watch
- Content Producer
- Social Scheduler
- Release Notes Writer
- Codebase Guide
- PR Describer
