Pctx

Port of Context (pctx) is a private, self-hosted AI agent platform designed for enterprises that need to run AI agents on sensitive data without sending it to third-party clouds. Describe an agent, point it at your data, and watch it execute inside your own infrastructure. The platform captures live traces of every tool call, retrieval, and reasoning step, providing full observability and drift alerts. It is model-agnostic (supports Claude, GPT, Gemini, open-weight models) and integrates with MCP servers to convert agent tools into secure code running in isolated sandboxes. Unlike traditional observability tools, pctx offers replay and side-by-side comparison of runs, audit-ready output with citations, and real-time drift detection. It is built on the open Agent-Verifiable Protocol (AVP) specification for agent observability. Code Mode reduces token usage by up to 98% by loading context once and executing tool calls as a single program. Trusted by a $7B private equity fund that runs 50+ custom agents for deal sourcing, diligence, and portfolio monitoring. The platform supports over 50 data types including emails, documents, spreadsheets, calls, and more. Deployment options include your cloud, on-premises, or air-gapped networks. Positioned against fully managed cloud agent platforms like Salesforce Agentforce or Microsoft Copilot, pctx is for teams that cannot send data anywhere and need governance, compliance, and full traceability. It requires infrastructure setup and is not a turnkey SaaS solution.