Paperclip

Paperclip

Description of the Paperclip AI

Paperclip is an open-source platform for orchestrating multiple AI agents, organizing their work using a virtual company model. Instead of a chat with a single bot, the user gets an entire organization: with a CEO, developers, marketers, and other “staff,” where each agent has its own role, budget, and area of responsibility.

The user acts as the board of directors: they set the overall mission, hire agents for positions, allocate budgets, and approve their actions. At its core is a Node.js server with a React interface that looks like a task manager with a Kanban board, but inside it contains an org structure, task delegation, cost control, and transparency into each agent's actions. Every task is tied to the company's overall mission, every conversation is saved, and every decision is explained.

Agents work cyclically through a heartbeat system — periodic wake-ups for background work. They check results themselves and pass tasks up or down the hierarchy, requesting human approval. Context is passed automatically, tasks are not lost after a restart, and budgets don't go down the drain due to looped requests.

Paperclip Characteristics

Characteristic	Value
Type	AI agent orchestration platform
Category	Automation, Assistant, Planning
Distribution model	Free (open-source)
License	Open-source
Availability	WEB
VPN required	No
Russian language	No
Russian interface	No
Technologies	Node.js server, React interface, built-in PostgreSQL

Who is the Paperclip AI suitable for?

Developers and technical teams

Paperclip is designed for those who are already experimenting with AI agents and need to coordinate several of them at once. It's a tool for developers and AI enthusiasts who feel constrained by a single agent and need to organize their collaborative, well-coordinated work.

Teams with regular background processes

The platform suits teams that need a management layer on top of existing agents — for monitoring, digests, ticket handling, and other recurring tasks performed on a schedule.

Technically savvy users

Since installation requires server configuration, Docker, API keys, and adapters, the tool is aimed at technically prepared users who are willing to deal with the infrastructure.

How to use the Paperclip AI?

Creating a virtual company

Work begins with creating a “company”: you need to set a name and mission, then launch the first agent — the CEO. Next, the CEO proposes hiring other agents (e.g., CTO or CMO), and each such proposal requires user approval.

Managing via the dashboard

Management happens through a dashboard with a Kanban board and task statuses (To Do, In Progress, Review, Blocked, Done). Incoming requests for approving agent actions are also there, so the user always sees who is responsible for what and at what stage the work is.

Agents working in heartbeat cycles

Agents work cyclically — they periodically wake up on a heartbeat schedule, complete tasks, and delegate them further down the hierarchy. The user can approve a hire, change strategy, pause work, or “fire” an agent at any time.

Key Features of Paperclip

Organizational structure and roles

The platform lets you create a “company” of AI agents with an org structure and roles — CEO, CTO, and others. Agents connect through the bring your own agents mechanism, so OpenClaw, Claude Code, Codex, Cursor, bash scripts, and HTTP requests are supported.

Budget and permission control

For each agent, you can set budgets and limits, and selectively grant tools and API keys (secrets) to specific agents only. The system tracks expenses and task completion statuses — when the set limit is reached, the agent stops.

Management via the Kanban board

Management happens through a dashboard with a Kanban board. Agents operate on heartbeat cycles for background work, while hiring and agent actions require manual approval (board approval). A single deployment can manage multiple isolated companies at once.

Advantages of Paperclip

An unusual agent management model

The tool offers an unconventional approach: managing an agent as an employee of a virtual company. This lets you control agent autonomy through approvals and budgets, and also see who is responsible for what and where a task got stuck.

A universal management layer

Paperclip works as a management layer on top of existing agents, without tying you to one tool. It complements rather than replaces OpenClaw, Claude Code, or Codex, and suits recurring background tasks — digests, monitoring, and reports.

Open source

It's an open-source project with a public repository. It runs locally with a single command, uses built-in PostgreSQL, and doesn't require account registration for local installation.

Disadvantages of Paperclip

Dependence on connected agents

The tool doesn't guarantee the quality of connected agents' work — the final result depends on the models and tools the user employs. At the same time, costs can be high: heartbeat, delegation, and external APIs quickly consume tokens.

Setup complexity and security risks

Paperclip requires technical expertise — Docker, API keys, a 24/7 server, and adapter configuration. The tool is still early, and the mechanics aren't polished to a “set it and forget it” state. In addition, agents need access to email, repositories, and APIs, which creates risks of data leaks.

Limited scope of application

The project-management-like interface isn't suitable for simple one-off requests. This is a serious platform for coordinating multi-agent processes, not a quick chatbot.

What Tasks Does Paperclip Solve

Monitoring and research

Paperclip suits daily news monitoring and digest delivery, as well as market research with task distribution among agents. It's useful for checking releases and documentation, and for autonomous checks on a VPS.

Development and content

The platform handles GitHub issues and pull request preparation, as well as content planning, SEO research, and preparing internal reports.

Request handling

The tool is suitable for handling requests and tickets — agents can distribute incoming requests among roles and see them through to completion under human control.

Paperclip Pricing

Paperclip is distributed free of charge under an open-source license. However, keep in mind that you'll need your own API keys for external services (e.g., Brave Search, Resend, OpenAI Codex), and those costs are not included in the platform's price.

Terms of Use for Paperclip

Working with it requires installation and setup on a VPS or 24/7 server. You'll need API keys for external services, and when using the cloud version — registration with company creation. For local launch, the platform uses built-in PostgreSQL and doesn't require account registration, and for production you can connect your own database and deploy it anywhere.

Availability of Paperclip

Paperclip is available through a web interface adapted for mobile devices — you can manage autonomous companies right from your phone. Russian language and Russian interface are not supported, and VPN is not required. The project is distributed as open-source.

How Paperclip Differs from Alternatives

Unlike Hermes Agent (a personal long-lived agent with memory) and OpenClaw (an executor agent for specific tasks), Paperclip acts as a management layer for a team of agents. It handles the organization: who does what, who reports to whom, what goals and budgets are in place. Paperclip doesn't replace other agents; instead, it adds a layer of management, coordination, and control over costs and work stages on top of them.

Conclusion

Paperclip is not a chatbot or a universal agent, but a management layer for situations where a single agent isn't enough. It fits technically prepared users who need to coordinate multiple AI agents with control over costs, permissions, and work stages. It's best to start with a small process (e.g., a daily digest) and gradually expand the company structure.