Runs on your machine • Works with any AI model

Your AI shouldn't redo the same work every single day

Ask an assistant to do a job and it does it once, then forgets. Vibe Engine has the AI write the job down as one readable file. After that it runs on your own machine, on a schedule, without the AI being involved at all.

See it doing real jobs Fix a problem you're having →
1 Binary
No Python, no virtualenv, no install tree
Any Model
Local or cloud — your choice, per step
Readable
See every file and URL before it runs
Your Machine
Runs locally, on your schedule

The other tools we build

Vibe Engine is the main one. These are the rest, all built the same way: local-first, no cloud lock-in.

High-Throughput Routing

Janus API Gateway

A native Go LLM router that ships as one static binary — no virtualenv, no pip tree. Multiplexes local Vulkan/Ollama runners and cloud models behind a single OpenAI-compatible endpoint.

  • Universal OpenAI /v1/chat/completions endpoint
  • Compiled goroutine concurrency — no Python GIL
  • Single static compiled binary — no pip, no venv
  • Local token budgeting & prompt deduplication
Inspect the Gateway →
Autonomous Execution Layer

Vibra-Ingenn (Vibe Engine)

The Unix pipe for the local LLM era. The AI writes the recipe once, then steps back. Native Go runtime executes workflows deterministically — no ongoing token cost on pure-execution recipes.

  • Single-pass recipe compilation — the model touches the task once
  • Pure Go runtime — no Python, no pip, no virtualenv
  • Unix philosophy: turns small local models into precision tools
  • Deep template resolution: machine-portable with {{desktop_dir}}
Explore Vibe Engine → Our Way vs Their Way →
Autonomous Video Automation

Vibe-Cut

An agent-driven video editor. You describe the cut in plain English, an AI writes a JSON shot list, and FFmpeg renders it — no timeline, no mouse, no editor.

  • Frame-accurate FFmpeg trimming from a JSON shot list
  • Contact sheets an AI can read to pick the cuts
  • 25+ natural local TTS voices — no cloud, no per-word billing
  • Whisper transcription, scene detection & burned-in captions
See the Render Pipeline →
Web Protocol & Ingress

Agent-Ready Modernization

Converting existing websites into Web-MCP enabled, agent-friendly sites. Unlocks discovery, GEO optimization, and A2A autonomous commerce.

  • Web-MCP server installation (/.well-known/mcp.json)
  • Generative Engine Optimization (GEO) & llms.txt
  • A2A autonomous agent transaction cards
  • Mobile Human Checkpoint relay architecture
Explore Agent-Ready Services →

Get Vibra-Ingenn (Vibe Engine PRO)

Download first. Your full 7-day trial starts automatically after installation.

Community / Free Edition

Starter Edition

$0 / free tier

The full deterministic recipe engine, running entirely on your own machine against your own models.

  • Core Deterministic Execution Engine
  • Local Model Support (Ollama / LM Studio)
  • Multi-Step Recipe Execution
  • Local-only: outbound API calls require PRO
Download Free (.exe)
★ Recommended for Power Users

Vibe Engine PRO

$8.99 / month after the optional trial

Everything in the free tier, plus recipes that reach beyond your own machine: web requests and cloud AI models.

  • Everything in Starter, plus:
  • Outbound web requests: call APIs, fetch pages, send alerts
  • Cloud AI models inside your recipes
  • Mix local and cloud models in one recipe
  • License key emailed to you after checkout
⚡ Subscribe to PRO ($8.99/mo)