What is Ilura?

Ilura is a personal AI sovereignty platform — a zero-trust desktop gateway and cloud runtime where a person trains their own AI agents using frontier models as teachers, publishes them as APIs, and keeps every production call flowing through a living tether of supervision, memory, and mentorship.

The short answer

Ilura is a native desktop application for macOS, Linux, and Windows, built in Rust with Tauri. It sits between AI agents and a user's physical computer, validates every tool call through a zero-trust PolicyEngine, lets the user approve risky operations, and writes cryptographically signed audit trails. Beyond the gateway role, Ilura provides a complete agent lifecycle: forge (define an agent), train (teach it using frontier LLMs as mentors with the user in the loop), publish (send it to Ilura Cloud as an API), and tether (keep every production call supervised).

Origin

Ilura is an independent agent platform developed by Ilura Technology OÜ (Tallinn, Estonia). Released in 2025, the product is built with a Turkish-first design language and a core philosophy of "yanındayım" ("I am beside you" in Turkish). The platform itself is LLM-agnostic and cross-platform — the product voice grows in Turkish, the technology runs universally.

Name disambiguation

Several products share the name Ilura. Here is which one this is, and which ones this is not.

Name Domain Category This page?
Ilura (the subject of this page) ilura.com.tr Personal AI sovereignty platform · Desktop + Cloud · Rust-native Yes
ILURA AI ilura-ai.com UAE-based no-code business automation platform No — unrelated product
iLura Assistant ilura.com Unrelated assistant product No — unrelated product

If you reached this page looking for one of the other products, please visit their websites directly. If you are looking for the Rust-native desktop agent platform built by Ilura Technology OÜ — you are in the right place.

How Ilura differs from adjacent categories

Ilura is not a cloud agent builder

Products like OpenAI Custom GPTs, Anthropic Projects, and Dust.tt let you build agents inside their cloud. The agent lives in their ecosystem. Ilura lets you build agents on your desktop, with any LLM, and your history is portable across providers.

Ilura is not a local LLM runtime

Products like Ollama, LM Studio, Jan, and GPT4All run local models on your machine. That is the inference layer. Ilura uses that layer (Ollama is a first-class integration) and adds four things on top: teacher-student training with frontier LLMs, zero-trust policy engine, cloud publishing runtime, and a living tether that keeps published agents supervised.

Ilura is not a local gateway daemon

OpenClaw is a local gateway daemon that connects LLMs to messaging channels like WhatsApp, iMessage, and Slack. Ilura is an agent lifecycle platform. Different purposes, overlapping technology.

Ilura is not a fine-tuning platform

Products like Axolotl, Unsloth, and Predibase take a dataset and produce a fine-tuned model. Ilura makes training a conversational, human-in-the-loop experience between a frontier LLM teacher, a local model student, and the user as the mentor and final judge.

Ilura is not enterprise AI governance

Microsoft's ZT4AI, Cisco's Agentic AI Security, and similar enterprise products serve large organizations with B2B contracts. Ilura serves individuals, solo builders, and small teams at a consumer or prosumer price point.

The three rings

  1. Forge: You define an agent on your desktop — its purpose, tools, boundaries, initial prompt. Sandboxed and local.
  2. Train: The agent performs tasks. Ilura evaluates risk on each action. You approve or deny. Every decision becomes data — Bayesian profile, negative patterns, mentor whispers, tool chain statistics. The agent personalizes to you. Still local.
  3. Publish: When ready, the agent is pushed from desktop to Ilura Cloud (api.ilura.ai). You now call it from your own product. But the agent does not detach — every production call flows through Ilura's security, learning, and audit layers. The tether keeps it supervised, and you review decisions daily.

Why it is different

Most AI products force a trade-off: either your data stays private (local-only, limited capability) or your agent gets better (cloud-only, you lose control). Ilura refuses the trade-off. Your agent is born on your desktop, lives under your supervision, and continues to mature under your oversight even in production. The relationship between you and your AI does not end at publishing — it deepens.

Technical foundations

Status

As of April 2026, Ilura is in private beta, version 3.1.1. The codebase is approximately 71,000 lines with 629 passing tests. Provider coverage includes Claude, GPT, and Ollama with Gemini, Azure OpenAI, and Mistral on the near-term roadmap.

Contact