Claude, GPT-4, Gemini — PaperclipAI Runs Them All Through Adapters
PaperclipAI doesn't lock you into a single AI provider. The platform is model-agnostic — each agent connects to an AI runtime through an adapter, and you choose which adapter (and therefore which model) each agent uses. Different roles can use different models based on what they're best at.
Built-in adapters
| Adapter | Runtime | Best For |
|---|---|---|
| claude_local | Claude Code CLI | Complex coding, analysis, nuanced tasks |
| codex_local | OpenAI Codex CLI | Code generation and editing |
| gemini_local | Gemini CLI | Multi-modal tasks, Google ecosystem |
| opencode_local | Multi-provider support | Flexible provider/model combinations |
| cursor | Cursor CLI | IDE-integrated development |
| http | Webhook delivery | External services and custom agents |
| process | Shell commands | Arbitrary command execution |
You bring your own API keys
Unlike SaaS AI tools with fixed subscriptions, PaperclipAI requires you to provide your own API keys for each provider. Your Claude adapter needs an Anthropic API key. Your Codex adapter needs an OpenAI key. This means your costs are based on actual token usage at provider rates — no markup, no middleman.
Mixing models in one organization
A powerful pattern: assign different adapters to different roles based on their strengths. Your CEO agent might use Claude Code for strategic thinking and complex analysis. Your engineering agents might use Codex for code generation. Your marketing agents might use Gemini for multi-modal content. Each agent's adapter is configured independently.
Creating custom adapters
If the built-in adapters don't cover your needs, PaperclipAI supports creating custom adapters. Any runtime that can call an API can serve as an agent. The adapter interface requires three modules: server (execute logic), UI (output rendering), and CLI (terminal formatting). This makes PaperclipAI extensible to any AI system, current or future.
Keep reading
ChatGPT Is a Chat Window. PaperclipAI Is a Company Operating System.
ChatGPT answers questions one at a time. PaperclipAI runs an entire organization of AI agents — with org charts, budgets, task management, and governance. Different tools for different problems.
PaperclipAI Is Free. Your AI Provider Bill Is Not. Here's the Real Cost Breakdown.
PaperclipAI is open-source — the platform itself costs nothing. Your costs are AI provider API usage (tokens), hosting, and your time. Here's what the real spending looks like.
Want to talk about what AI can do for your business?
30 minutes, no sales pitch. Just an honest conversation about where AI can move the needle for you.
Book a Free Call