patty

What patty does

Patty Co., Ltd. (주식회사 패티) is a privately held Korean AI software company founded in 2025 that builds an enterprise AI platform in which models, GPU infrastructure, and AI agents are connected through a single governance layer rather than delivered as standalone model APIs. Its product portfolio comprises five flagship products: Patty Code, a Korean-first AI coding agent shipped as a single Go binary for macOS, Linux, and Windows; PCCP, the AI development control plane with three deployment profiles (Cloud, Enterprise, Sovereign); Ergazo, an AI-native work execution platform enforcing a propose-approve-execute-record workflow; Crew, a human-plus-agent shared workspace built on the Nostr protocol with Schnorr-signed event logs; and Pilot, a pre-release v0.1.0 control plane for AI agent organizations with org charts, per-agent budgets, and approval gates. All five products share the DARI (Delegated Authorization and Receipts for Inference) protocol for cryptographic receipt integrity using CBOR records and COSE-Sign1 signatures.

Underlying technology spans 21 distinct capabilities divided across models and training (pre-training and continued pre-training, SFT pipelines, GRPO/RLHF alignment, dataset generation, Korean-specialized modeling, custom model development), evaluation and benchmarks (benchmark engineering, model evaluation and red teaming), inference and infrastructure (distributed GPU serving, KV cache architecture, wide-EP MoE serving, multimodal inference, RL rollout acceleration, GPU farm operations), and applied AI (Korean-specialized RAG, document intelligence, LLM-native browser, agent orchestration, AI workforce transformation, ViewSpec rendering, on-premises deployment). The platform ships 12 runtime adapters spanning Claude, Codex, Cursor, Gemini, Grok, Kimi, Opencode, Pi, Hermes, and OpenClaw Gateway, plus 7 sandbox providers including Kubernetes, Daytona, E2B, Modal, Cloudflare, exe.dev, and Novita; tools and data are exposed through the Model Context Protocol, and Crew bridges to external agents over ACP/JSON-RPC.

The company pursues a sales-led, consultative go-to-market targeting regulated Korean enterprises — healthcare, financial services, manufacturing, and public sector — where verifiable AI control, data residency, and audit-grade evidence are architectural requirements rather than policy aspirations. Revenue mechanics are enterprise platform licensing and deployment-profile-based fees; pricing is not publicly disclosed and the company explicitly declines to publish figures without measurement context and verification. No revenue, customer logos, headcount, or funding rounds were disclosed in the source materials, and Pilot in particular ships as a self-hosted pre-release, indicating the company is at an early commercial stage with infrastructure investment and product buildout still in progress. The founding leadership team (Patrick Rho, David Shin, Iila Kim, Leo Kwon) brings systems-level AI and operational governance experience from Google, Amazon, LG CNS, NHN, and prior corporate strategy roles.

Name

patty

Legal name

Patty Co., Ltd. (주식회사 패티)

Website

https://patty.io

Company type

Private

Founded year

2025

Operating status

Operating

Short description

Patty Co., Ltd. is a Korean AI company building an enterprise AI platform with five products (Patty Code, PCCP, Ergazo, Crew, Pilot) unified by the DARI cryptographic governance protocol, and tailored to Korean-language modeling and Korean data sovereignty requirements for regulated enterprise buyers.

Ownership category

akta.pro rank

Product category

Enterprise AI Governance Platform

NAICS

Software Publishers (51321), Computer Systems Design and Related Services (5415)

SIC

Services-Prepackaged Software (7372), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)

akta.pro secondary industries

Responsible AI, Security & Privacy Platforms (Safety, Guardrails, PII) (HDAEANAG), AI Application Enablement Platforms (Copilot/Agent Frameworks, SDKs) (HDAEANAJ), Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem) (HDAEANAC)

Keywords

• Enterprise AI platform
• AI governance software
• AI coding agent
• Agent orchestration platform
• AI control plane

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Infrastructure, Operations

Revenue model

1. Enterprise AI Platform Licensing: Patty offers enterprise AI products (Patty Code, PCCP, Ergazo, Crew, Pilot) positioned for organizations requiring verifiable AI control and governance. Based on the consultative sales motion and enterprise targeting, revenue is derived from software licensing and platform fees. The company is pre-public launch with security/compliance frameworks under development.

Go-to-market motion2 records

Distribution channels2 records

Marketing channels6 records

Core offering

Patty builds an enterprise AI platform that connects models, GPU infrastructure, and AI agents through a single verifiable governance layer (the DARI protocol and PCCP control plane). Its five flagship products — Patty Code, PCCP, Ergazo, Crew, and Pilot — provide a propose-approve-execute-record workflow that makes every AI action visible, controllable, and provable through cryptographic receipts. The platform supports Cloud, Enterprise, and Sovereign deployment profiles, including air-gapped on-premises setups for Korean organizations requiring data residency.

Product overview

Patty offers a suite of five flagship products built on a shared governance architecture centered on the DARI (Delegated Authorization and Receipts for Inference) protocol. The portfolio comprises: Patty Code (Korean-first AI coding agent), PCCP (AI development control plane), Ergazo (AI-native work execution platform), Crew (human+agent collaboration workspace on Nostr), and Pilot (control plane for AI agent organizations). All five products share the same control principles — proposal, approval, execution, and evidence recording — unified under the DARI protocol and PCCP governance layer. The stack spans from coding agents and individual work execution through multi-agent collaboration and organizational-level agent governance.

Differentiator

Problem solved

Functional benefit

Products and services

• Patty Code Korean-first AI coding agent that enables developers to converse in Korean, execute tools, and complete coding tasks from a single Go binary running on macOS, Linux, and Windows. All actions execute under PCCP control plane policies and approvals.
• PCCP (Patty Code Control Plane) AI development control plane that makes all AI development actions visible, controllable, and provable. Governs human and agent identity, repository scope, command allowlists, and approval for exported artifacts. Provides three deployment profiles: Cloud, Enterprise, and Sovereign.
• Ergazo AI-native work execution platform where AI proposes and a human approves before execution. Manages the propose-approve-execute-record workflow for knowledge work tasks.
• Crew Human + AI agent shared workspace 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. Agents are citizens with keypairs, not guests.
• Pilot Control plane for AI agent organizations — the control tower where a human board sets company goals, hires AI agents into an org chart, gates spend and approvals, and tracks work, cost, and progress. Pre-release v0.1.0, self-hosted deployment.

Quantifiable outcome

• Semantic snapshot generation ~14-16ms regardless of page size (LLM-native browser, debug build, small fixtures)
• +2 more outcomes

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration21 records

AI capability16 records

Feature43 records

Scale indicators14 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Naver Hyperclova: Naver's HyperCLOVA X is a Korean-specialized large language model and enterprise AI platform with strong focus on Korean language quality and data sovereignty. Most directly comparable to Patty as a Korean-first AI platform serving regulated Korean enterprises.
• LG AI Research (EXAONE): LG AI Research develops EXAONE, a Korean-specialized foundation model family, with enterprise applications and governance focus. Comparable to Patty as a Korean AI specialist with both model and platform ambitions serving LG-affiliated enterprises and external customers.
• Upstage: Korean AI company developing document intelligence, OCR, and large language models with enterprise focus. Comparable to Patty as a Korean AI company with strong document understanding and enterprise deployment capabilities, including sovereign deployment options.

Broad incumbents

• Anthropic: Enterprise AI platform with strong emphasis on safety, governance, and Constitutional AI principles. Comparable to Patty in its focus on responsible AI, verifiable safety, and enterprise-grade AI governance, but at vastly larger scale and broader market.
• Cohere: Enterprise AI platform focused on customizable, secure, and deployable language models for regulated industries. Comparable to Patty in targeting enterprise customers with data sovereignty and on-premises deployment requirements, though with broader geographic focus and larger scale.
• Together AI: AI infrastructure platform offering distributed GPU serving, fine-tuning, and inference with focus on open models and customization. Comparable to Patty's distributed GPU serving and inference capabilities, including MoE serving and KV cache architecture, though at larger scale.
• Snorkel AI: Enterprise AI platform focused on data-centric AI development with programmatic labeling, evaluation, and governance for regulated industries. Comparable to Patty in serving enterprise customers with strong data governance, audit, and customization requirements.

Regional players

• Sakana AI: Japan-based AI research company developing language-specialized foundation models with focus on Japanese language quality. Comparable to Patty as a non-English AI specialist serving a specific linguistic market with sovereign considerations, though focused on Japanese rather than Korean.

Emerging players

• CrewAI: Open-source multi-agent orchestration framework with enterprise deployments. Comparable to Patty's agent orchestration capabilities in Pilot, focusing on multi-agent workflows with role-based agent coordination, but without the governance and control plane layer.
• Cursor: AI-native code editor with agentic coding capabilities and broad developer adoption. Comparable to Patty Code as an AI coding agent with terminal-based workflows, but focused on the developer experience rather than enterprise governance and audit.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat7 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/patty-io

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about patty