About LetKite
LetKite, the execution layer between AI and machines, lets the AI people already use act on real servers, factory equipment and homes — safely, and on record.
Why we built LetKite
Robots are how AI acts in the physical world. For the digital world — the servers, production lines and home networks that run everyday life — AI still has no hands. Models can explain exactly how to fix a failing server, and then a person has to open a terminal and type it in.
Organizations already pay for capable AI. What is missing is a safe way for it to touch real systems: one that grants access explicitly, asks for approval before sensitive actions, and keeps a record that stands up to an audit. LetKite is that layer.
What we believe
- Model-neutral. LetKite works with Claude, Kimi, GLM and other compatible AI. You choose the model; LetKite does not tie you to a single vendor.
- No new AI spend. LetKite connects to the AI plan you already have through its standard connector feature.
- Control before convenience. Permissions, approvals and audit come first; every feature is built on top of them.
- Open at the core. LetKite Lite is MIT licensed. Anyone can inspect, run and improve the part that touches their machines.
Why a kite
A kite goes where the wind takes it, but only as far as the string allows. In LetKite, the AI is the kite, your machines are where it flies, and the gateway holds the string — you decide how long it is.
Where we are
LetKite is an independent team based in Singapore. We offer our service in multiple languages, and the product works with many AI models.
Independent
LetKite is not affiliated with or endorsed by OpenAI, Anthropic, Moonshot AI, Zhipu AI or OpenClaw. Product names appear on this site only to describe compatibility.