AI agent company orchestrator
exployt vs Paperclip
Paperclip is a free, MIT-licensed, self-hosted control plane that organises agents you already use — Claude Code, Codex, Gemini CLI, Cursor, Hermes, OpenClaw — into an AI "company" with an org chart, heartbeats, budgets and board approvals.
Different goals
Make one person as powerful as a whole software company: a managed team of AI agents that plans, builds, tests and debugs production-grade software on their own machine.
Let one operator run a goal-driven, largely autonomous business staffed by bring-your-own AI agents, with company-style governance and cost limits.
What you can expect with exployt
- Start from a desktop app with its own local database — no server to host or maintain.
- Work with engineering tools where the work happens: debugger, Roslyn diagnostics, profiling, test runner and UI automation.
- Get memory built in: learnings, decisions, an error database, skills and a wiki with local semantic search.
- Drive Claude Code and Codex and call ~200 AI model & tool providers directly — 1,500+ models, including free local Ollama models.
exployt vs Paperclip — side by side
| What matters | exployt | Paperclip |
|---|---|---|
| Built for | One person shipping serious software | Running goal-driven AI-agent companies |
| Primary focus | Building and maintaining software | Business operations across engineering, marketing, sales and ops |
| How it runs | Windows desktop app with a local database (macOS in progress) | Self-hosted Node.js server + browser UI (macOS, Linux, WSL2, Docker) |
| Native Windows | Yes | Not documented (WSL2 or Docker) |
| Agent runtimes | Claude Code, Codex and Qwen Code + direct API agents | 12 adapters incl. Claude Code, Codex, Gemini CLI, Cursor, Hermes, OpenClaw, shell, HTTP |
| Direct model calls and local models | Built in: ~200 AI model & tool providers, 1,500+ models, incl. free local (Ollama) and image & video | None of its own; local models only through an agent such as OpenCode or Hermes |
| Team model | AI managers (CTO/CPO/PM) with spawn rights and approval tiers | Org chart (CEO, CTO, managers, workers) with reporting lines and delegation |
| Human oversight | Approval tiers, plan approval modes, permission rules | Board approvals, review stages, escalation after repeated rejections |
| Scheduling | Routines (time, interval and event triggers) + inactivity wake-ups | Heartbeats + cron, webhook and API routines |
| Persistent memory | Built in: learnings, decisions, error database, skills, wiki + semantic search | No built-in durable memory yet (plugins) |
| Real debugging | Debugger with breakpoints (.NET), browser debugging, Roslyn diagnostics, profiling | No debugger integration documented |
| AI cost | Up to 70% lower AI costs; bring your own keys or run free local models | You pay each agent's own model usage |
| Budget limits | Token budgets on targets; estimated cost per agent, task and feature | Per-agent budgets with hard stop and auto-pause |
| Users and companies | One user; seats bound to a device | Several users with roles; unlimited isolated companies |
| License and price | Proprietary; free for private use, paid plans for commercial use | Open source (MIT), free; you pay your agents' model usage |
Why exployt wins
- Built for software: debugger, diagnostics, profiling, tests and per-agent change tracking inside the orchestrator.
- Memory and knowledge built in, with local semantic search — no plugin needed.
- Calls ~200 AI model & tool providers directly, including free local models, and also drives Claude Code and Codex — all from one Windows desktop app.
Where Paperclip is strong
- Free and MIT-licensed, with a large, fast-moving community.
- Works with the agents you already use through a dozen adapters — including Hermes and OpenClaw.
- Mature company-style governance: org charts, board approvals, hard budget stops and isolated companies for several users.
The verdict
Pick Paperclip to run a goal-driven AI "company" across business functions with agents you already use, strict budgets and multi-user governance — on your own server. Pick exployt when the company you are building is a software product: a desktop team of AI agents with memory, debugger, diagnostics and tests built in.
Frequently asked questions
Is exployt a good Paperclip alternative?
If your agents mainly build software, yes. Paperclip coordinates bring-your-own agents across a whole business and leaves the engineering to those agents; exployt is a software-development platform where AI managers, memory, a debugger, diagnostics and tests are built in.
Does Paperclip run on Windows?
Paperclip's official install guides cover macOS, Linux and WSL2, plus Docker; a native desktop app is on its roadmap. exployt is a native Windows desktop app today, with macOS in progress.
Can I use local models with Paperclip?
Paperclip does not call models itself — its adapters run agents such as Claude Code, Codex or Hermes, so local models only work through an agent that supports them. exployt connects to Ollama directly, next to the cloud providers.
Ready to lead your AI army?
exployt exists to make one person as powerful as a whole company: run many well-organized AI agents in parallel to build, test and debug production-grade software — all on your own machine, with your choice of AI models.