Overview
Nelly is a no-code AI agent platform that enables users to author, run, and orchestrate teams of specialized agents using natural language descriptions. Users create agents, equip them with tools (browser, database, etc.), and break complex workflows into sub-agents that collaborate. The product emphasizes conversational authoring, thread-based context, department-style organization via drag-and-drop, and rich observability (tool usage, browser screenshots, step-by-step progress). Agents execute locally on the user’s machine while LLM computation runs in the cloud; data is stored locally with only conversation history sent to LLM providers. Nelly offers integrations with external LLM providers (Anthropic, OpenAI, OpenRouter, LM Studio) and support for Model Context Protocol (MCP) servers for custom tools. A marketplace for publishing and monetizing agents/tools is planned.
Key Features
No-code Agent Authoring
Create and configure AI agents using natural language descriptions and structured instructions without writing code.
Agent Orchestration & Teams
Compose agents into departments and agent teams; delegate tasks to specialized sub-agents to break down complex workflows.
Tool Integrations & Connectors
Equip agents with tools such as a browser, database adapters, and custom tools via MCP; agents can interact with external services.
Observability & Audit Trails
Detailed activity insights including tool usage logs, browser screenshots, and step-by-step progress to monitor and debug agents.
Local Execution & Data Privacy
Agents execute locally on the user’s machine and data is stored locally; only conversation history is sent to the chosen LLM provider.
Custom LLM Provider & MCP Support
Connect external LLM providers (Anthropic, OpenAI, OpenRouter, LM Studio) or use Model Context Protocol (MCP) servers for custom tools and unlimited messages when self-hosting the LLM.
Who Can Use This Tool?
- Non-developers:Build and deploy automation agents using natural language without coding.
- Product & Operations Teams:Automate repetitive workflows and orchestrate multi-step processes across tools.
- Developers & AI Engineers:Integrate custom LLM providers, use MCP for custom tools, and extend agent capabilities.
Pricing Plans
Start building and using AI agents for free.
- ✓Unlimited AI agents
- ✓30 messages per 30 days (rolling window)
- ✓No credit card required
When you need more control or higher message limits.
- ✓Unlimited AI agents
- ✓100 messages per 30 days (rolling window)
- ✓Unlimited messages when using your own LLM provider
- ✓Priority support and early access
For power users who need even higher message limits.
- ✓Unlimited AI agents
- ✓250 messages per 30 days (rolling window)
- ✓Unlimited messages when using your own LLM provider
- ✓Priority support and early access
Pros & Cons
✓ Pros
- ✓No-code agent authoring lowers the barrier for non-engineers to build complex automations.
- ✓Local agent execution and local data storage improve privacy and control (LLM in cloud; local stored data).
- ✓Flexible LLM support: default uses Anthropic Claude Sonnet 4 but supports connecting external LLM providers and MCP servers.
- ✓Strong observability: step-by-step progress, tool usage logs, browser screenshots for auditing and debugging.
- ✓Marketplace planned to enable publishing and monetization of agents and tools.
✗ Cons
- ✗Product is in beta; features, pricing, and limits may change.
- ✗Message-based limits may be restrictive for heavy workloads (and there’s no message top-up in beta).
- ✗Features and documentation (features page returned a 404 in one crawl) appear incomplete in places.
- ✗Limited public adoption / third-party reviews available at time of review.
- ✗Enterprise pricing and detailed SLAs not publicly listed.