Propose
AI proposes the next action within the work context and policy boundary.
Verifiable enterprise AI
Patty builds every AI your enterprise needs
Start your AI adoption with Patty products.
Starts with a proposal, ends with a record
Patty separates proposal, approval, execution, and recordkeeping.
Approve
An authorized person reviews the scope and impact before granting authority.
Execute
The control kernel runs only approved work through isolated tools and infrastructure.
Record
Action, rationale, result, and provenance remain as a verifiable receipt.
Different products, one principle
Each product handles different work while managing proposals, approvals, execution, and evidence through the same operating model.
Coding agent
Mirr Code (Personal/Team)
A coding agent that converses in Korean in your terminal, executes tools, and stays with the task until it is done. One Go binary runs identically on macOS, Linux, and Windows — and every action executes under the policy and approval of the Mirr Enterprise control plane.
EnterpriseMirr Enterprise (Company)
The coding agent, execution control layer, and audit evidence bundled into one enterprise agreement, layered on your identity, repositories, internal tools, and private models.
Government & public sectorMirr Gov (Government/Public)
A public-sector and sovereign bundle built around disconnected networks and air gaps, in-country data residency, local PKI and KMS, and an offline model catalog.
Work platformErgazo
Ergazo is an AI-native work execution platform for Korean companies. Employees ask in Korean or English from one governed workplace chat, read the live business data they are authorized to see, edit through safe tables and drafts, and apply changes under typed services, policy checks, approvals, and an append-only audit trail. AI proposes, people approve, systems execute — and everything is recorded.
Collaboration platformCrew
A self-hosted collaboration platform built on the Nostr protocol. Chat, forum, docs, git, workflows, and voice huddles — every action by every person and AI agent lands as a Schnorr-signed event in one log, and a single Rust relay is the source of truth for all of it.
Organization operationsPilot
The control plane where a human board sets the company goal, hires AI agents into an org chart, gates spend and approvals, and tracks work, cost, and progress from one dashboard. Pilot does not run your agents — agents run wherever they run and phone home on a heartbeat.
Verifiable technical capabilities
From model training to distributed inference and agent execution, we build every capability in-house. See how each one works.
Measurements and specifications
Security and governance are operating architecture
Training and warnings will not stop it. Policy has to decide on every request before data touches the model or infrastructure. Patty's LLM gateway makes that decision first and never lets unauthorized access through.
DARI Protocol
Receipts, not trust, for AI actions — the open delegated-authorization and receipt infrastructure.
Security Model
Defense-in-depth authorization across every AI action.
Audit & Provenance
DARI receipts, line-level human/AI provenance linked to Git, encrypted trace vaults.
LLM Gateway & Policy Engine
Server-side model authority; per-request policy decisions before infra contact.
DLP & Privacy
Input masking before retrieval/LLM contact; privacy-law-aligned handling.
Responsible AI
AI proposes, humans approve, systems execute, everything records.
Every layer of the AI stack, built in-house
Korean modeling, GPU serving, agent runtimes, organizational control. Patty builds the full enterprise AI system.
Send your requirements and we will propose a review scope.