The home of sovereign AI agents
Deploy AI agents with just a few lines of code. Secure. Sovereign. Only on the Internet Computer.
Start building
The Internet Computer allows you to build and run software services that are tamper-proof, giving you true ownership and digital sovereignty. We call them canisters. You can now access LLMs from within these canisters with just a few lines of code to build AI agents that you truly own and control.
Prompting
- Motoko
- Typescript
- Rust
import LLM "mo:llm";
await LLM.prompt(#Llama3_1_8B, "What's the speed of light?")
import * as llm from "@dfinity/llm";
await llm.prompt(llm.Model.Llama3_1_8B, "What's the speed of light?");
use ic_llm::Model;
ic_llm::prompt(Model::Llama3_1_8B, "What's the speed of light?").await;
Chatting with multiple messages
- Motoko
- Typescript
- Rust
import LLM "mo:llm";
await LLM.chat(#Llama3_1_8B, [
{
role = #system_;
content = "You are a helpful assistant.";
},
{
role = #user;
content = "How big is the sun?";
}
]);
import * as llm from "@dfinity/llm";
await llm.chat(llm.Model.Llama3_1_8B, [
{
content: "You are a helpful assistant.",
role: llm.Role.System,
},
{
content: "How big is the sun?",
role: llm.Role.User,
}
]);
use ic_llm::{Model, ChatMessage, Role};
ic_llm::chat(
Model::Llama3_1_8B,
vec![
ChatMessage {
role: Role::System,
content: "You are a helpful assistant".to_string(),
},
ChatMessage {
role: Role::User,
content: "How big is the sun?".to_string(),
},
],
)
.await;
AI Agent Ecosystem on ICP

Anda

ALICE

KinicUses Internet Identity
3,000,000+ searches

DecideAIUses Internet Identity
50,000+ inference requests
Pickpump
Sovereign AI roadmap
Ignition
Foundational LLMs
AI agents are a growing paradigm in software, and the Internet Computer is uniquely positioned to be the home of AI agents. Compared to the traditional IT stack, AI agents on the Internet Computer can securely and easily manage digital assets, transact amongst each other, be fully sovereign, and be tokenized and managed by DAOs. This milestone focuses on providing access to foundational LLMs, along with basic tools to make it as simple as possible to deploy agents on the Internet Computer.
LLM Canister MVP
A canister will be made available that provides an API for processing LLM prompts. This canister is an MVP and will initially rely on a centrally-managed service by DFINITY for processing the prompts. A small set of foundational models will be supported.
LLM Developer Experience
Tools and libraries to simplify the local development and the interaction with the LLM canister
LLM Workers
Introduce worker nodes that specialize in processing LLM prompts. Workers will have security guarantees similar to canister smart contracts on the Internet Computer