LOADING EXOSUITS
0/2 MODELS READY
The machine writes the diff. The human clicks approve. Lazyware adds a cheap bio-signal checkpoint before an agent gets permission.
open-source agent-control layer • $2-$3 sensorsDEVELOPER CONTROL, REAL BUILD
Developers are not just prompting AI anymore. AI agents are prompting developers to approve actions. Lazyware turns the lazy accept-click into a visible checkpoint: body signal, human attention, then permission.
First humans prompted AI. Now AI prompts humans: approve this diff, run this command, merge this branch. Lazyware puts a bio-signal checkpoint inside that loop.
FROM PROMPTING TO BEING PROMPTED
This is where it started: a working IDE, files, tabs, panels, and a developer actually driving the machine.
Then some genius glued the model straight into the sidebar. The IDE stayed, but the agent started sharing the wheel.
Then we somehow evolved backwards: from full IDEs to terminal screens, Claude maxxing, and approving agent actions in text.
Codex is a coding agent, but the cockpit is still mostly chat: prompt it, wait, inspect diffs, approve. You see patches, not a real control surface.
This is the actual destination: green computer interfaces, the vest, cheap sensors, buttons, eyes, muscles, and any signal that lets developers control agents faster than another chat box.
WHY RAISE $5K
Turn the prototype into a readable permission signal that AI agent tools can treat as a checkpoint.
Let people compare hands-free prompt methods instead of pretending one interface should decide the future.
Publish the code, diagrams, demo, and build guide so anyone can copy, criticize, or fork Lazyware.
Image sources: Visual Studio Code, OpenAI Codex, Claude Code.
BACKER CART
This is not a sealed product checkout. It is the pledge cart for funding the open-source Lazyware demo, build guide, and bio-signal approval layer.
Guide, demo, diagrams, source release, and the strange little approval checkpoint for agent actions.
SIGNAL REGISTRY
Names enter a private approval queue first. Approved names show on the public leaderboard. Bad names get stopped at the gate.