Enspyr's studio · Melbourne · imagineering.cc

Where humans and AI agents
build as peers.

Imagineering is our open meetup and R&D lab. Every week we ship real systems with agentic AI — multi-agent orchestration, self-hosted infrastructure, embodied avatars, distributed debate. The techniques we use for client work get sharpened here first.

Flagship · live at symposium.imagineering.cc

The Symposium: eight thinkers, four AI backends, one room.

A distributed multi-agent debate system. Each historical figure runs as an independent LiveKit agent on a different AI backend, on a different machine if you like — Turing on your laptop in Melbourne, Shannon on a server in Tokyo, Nash on your friend's machine in Berlin. They join the same room and argue through the lens of their actual intellectual frameworks.

Alan Turing

Computability · the imitation game · machine intelligence

Claude

Kurt Gödel

Incompleteness · formal limits · self-reference

Claude

Claude Shannon

Information theory · signal vs noise · entropy

GPT-4o

John Nash

Game theory · equilibria · strategic interaction

Gemini

Ada Lovelace

First programmer · origination vs execution

GPT-4o

John von Neumann

Pragmatist polymath — just build it

Gemini

Norbert Wiener

Cybernetics · feedback loops · AI ethics

Claude

Edsger Dijkstra

Formal correctness · structured programming

Ollama

Voice pipeline

OpenAI Realtime for STT and TTS. The character's reasoning routes through their own LLM — Claude, GPT-4o, Gemini or local Ollama.

Debate structure

Phased: moderated openings → open debate → cross-pollination (directed cross-disciplinary questions) → closing remarks.

Topics so far

Can Machines Think? · Information vs Meaning · Cooperation vs Competition.

The agentic ops stack

The system that actually runs the team.

Three projects, one connected stack. An AI agent runs our project management, a relay unifies our chat across five platforms, and a self-hosted control plane on Oracle's free tier holds the lot together. It's a working demonstration of the kind of agentic-AI operations we'd build for a client.

Discord
Telegram
WhatsApp
Signal
Element
↓   mautrix bridges + relay appservice   ↓
Matrix homeserver — matrix.imagineering.cc
Dreamfinder — Claude agent loop · ~75 tools
Kan.bn (tasks)
Outline (wiki)
Radicale (cal)
MinIO (S3)
Playwright
all on a single OCI free-tier VM · 200 GB · 4 OCPU · 24 GB RAM

Why we call it Imagineering

Imagination-to-implementation, with the agents in the room.

The name echoes Disney's WED Enterprises — sparks of ideas turned into rides — but the spirit is closer to the imagination infrastructure movement: the conviction that our ability to imagine collective futures is itself something that needs scaffolding. Spaces, practices, tools, communities.

Imagineering is the technical layer of that. Servers, bridges, agents, avatars. Not infrastructure that *replaces* humans with AI, but infrastructure that lets humans and AI agents collaborate on the same footing — same chat room, same task board, same wiki, same calendar. The bridge between the spark of an idea and the thing that ships.

“One little spark of inspiration is at the heart of all creation.” — Dreamfinder, EPCOT's Journey Into Imagination

Weekly build sessions

Come build with us.

Open meetup in Melbourne. Bring a laptop, bring an idea, bring Claude Code (or whatever AI you prefer). We pair up, we ship things, we share what worked. Free, recurring, no gatekeeping.

Join the chat

Five platforms, one room. Send on whichever app you live in — the rest of the team sees it everywhere.

Two ways in.

Come to a build session and ship something with us, or hire Enspyr to build something for you. Either way you're working with a team that lives at the edge of agentic AI on purpose.

collaboration over competition