Nelly
Build, test, organize, and use desktop AI agents through natural-language instructions without coding.
About Nelly
Nelly is a desktop AI agent platform for building, testing, organizing, and using custom agents without writing code. Studio turns natural-language instructions into agents. Builders can equip them with tools such as a browser and database, and supports iterative testing before use. Agents can delegate parts of a workflow to specialized sub-agents, while Chat provides threaded conversations and departments for organizing an agent team. Activity views expose tool use, browser screenshots, and step-by-step progress. Agents and their data run and remain on the local computer; conversation history and context are sent to the selected cloud LLM provider. Paid plans add custom provider connections for OpenAI, Anthropic, OpenRouter, and LM Studio. The Free plan includes 30 messages per rolling 30-day window, Pro includes 100, and Pro+ includes 250. Agent-to-sub-agent exchanges also count as messages, but Nelly limits do not apply when a paid user supplies a custom LLM provider; that provider’s own policies and limits still apply. The planned Marketplace and MCP server support are marked as coming soon, so they are not current workflow capabilities.
Best for
Best use cases
A recurring task needs an agent with browser or database tools and conversational testing before regular use.
A complex workflow benefits from splitting work among specialized sub-agents and tracing their activity step by step.
Multiple agent conversations need separation into threads and departments so ongoing work stays organized.
Compare before choosing
Choose Nelly when desktop-local agent data and a no-code builder matter more than a fully cloud-hosted workspace.
The free tier supports basic evaluation, while higher-volume use depends on paid message limits or a custom LLM provider.
Marketplace publishing and custom MCP server connections are planned features, not reasons to choose Nelly for a current deployment.
Features
Build custom agents
Describe an agent in natural language, add available tools, and refine its behavior through conversational testing.
Coordinate sub-agents
Divide a complex workflow into smaller jobs and delegate them to specialized sub-agents within an agent team.
Run organized agent chats
Use threads, departments, and activity details to keep conversations organized and inspect each agent step.
Pros
- Natural-language building and conversational testing reduce the setup cost for users without coding experience.
- Local storage for agents and their data gives desktop users more direct control over working files and configuration.
- Paid custom-provider support removes Nelly platform message limits for OpenAI, Anthropic, OpenRouter, and LM Studio usage.
Cons
- The Free, Pro, and Pro+ tiers cap platform usage at 30, 100, and 250 messages per rolling 30 days.
- Agent-to-sub-agent exchanges count separately, so delegated workflows can consume the included message allowance faster.
- Marketplace publishing and custom MCP server connections are still marked as coming soon and cannot support current workflows.
Pricing Plans
Free
- Unlimited AI agents
- 30 messages per rolling 30 days
- No credit card required
Pro
- Unlimited AI agents
- 100 messages per rolling 30 days
- Unlimited messages with a custom LLM provider
- Priority support and early access
Pro+
- Unlimited AI agents
- 250 messages per rolling 30 days
- Unlimited messages with a custom LLM provider
- Priority support and early access
* Pricing information is for reference only. Please refer to the latest announcement on Nelly's official website.