Back to products
WUPHF by Nex.ai

WUPHF by Nex.ai

AI employees who build their own knowledge base

Overview

What it is

WUPHF is a collaborative office of AI employees who build and maintain their own knowledge base to never lose context for the tasks you give them. Supports Claude Code, Codex, OpenClaw agents and local LLMs via OpenCode. Chat with your agents via TUI, Web or Telegram. Open source. Runs on your machine, with your keys.

Intent

I need it when

Mix multiple AI model providers in a single workflow without rebuilding integrations

WUPHF supports Claude Code, Codex, Hermes Agent, OpenClaw bridged agents, and local LLMs via OpenClaw Gateway HTTP. Each agent can run on a different provider (e.g., PM on Claude Opus, ENG on Codex) while collaborating in the same channel with unified @mention semantics.

Customize AI agent behavior and tooling to match internal processes and existing systems

Every agent is a JSON config file with system prompt and tool list. Users can read, edit, and fork the founding-team pack, swap in internal tooling, and wire up custom integrations without six-month implementation cycles.

Reduce SaaS subscription overhead and per-seat licensing costs while maintaining team coordination

WUPHF is MIT-licensed, free, and open source. It runs entirely local with no per-seat pricing, usage fees, or cloud backend required. Users can fork agent configs and integrate internal tooling in an afternoon without vendor lock-in.

Automate multi-step project workflows without manually routing tasks between tools and team members

WUPHF runs a local AI office where agents with defined roles (CEO, ENG, DSG, CMO, PM) automatically decompose goals, coordinate handoffs, surface blockers, and execute work 24/7 without human intervention in the loop. Users drop one sentence and walk away while agents argue dependencies and resolve issues independently.

Maintain persistent context and institutional knowledge across multiple projects without losing task history or decisions

WUPHF stores channel history in local SQLite, maintains per-agent notebooks, and promotes durable conclusions to a shared team wiki (git repo of markdown). Agents remember PR numbers, blockers, and decisions from previous days, eliminating context loss between sessions.

Drop

Not a fit when

  • User requires a fully managed cloud solution with hosted backend and zero local infrastructure setup
  • User needs real-time multi-user collaboration across distributed teams without private network setup (Tailscale/WireGuard required for local sharing)
  • User requires production-ready failure handling and guaranteed uptime; product is still iterating on failure modes
  • User cannot run local processes or prefers SaaS-only tools without command-line installation
  • User needs third-party integrations like Figma and Slack fully implemented; some are still natural-language placeholders
Commercials

Pricing

Free and open source. No per-seat pricing, no usage fees.