It is a local AI partner that runs on a Mac, maintaining a 'proprietary memory' of the user's work context and past judgments within the Mac to autonomously support daily tasks.

What is the biggest feature of Kocoro?

Unlike traditional AI, it has a 'proprietary memory' feature that eliminates the need to re-explain. Data is primarily stored within the Mac, and it uses an open-source design that allows users to verify what it is doing.

What kind of Mac tasks does Kocoro support?

It supports tasks called from chat apps like Slack, LINE, and Lark, providing responses and executing tasks based on project background, preferences, and past decisions.

Has Ptmind been in the AI business before?

For over 15 years, Ptmind has developed the web experience optimization platform 'Ptengine' and honed its customer understanding technology. Kocoro is a product that expands this understanding into 'judgment and action'.

What is the pricing model?

It is a fixed monthly subscription with a free tier. The pricing structure is transparent, with no unexpected additional charges.

Giving AI a 'Heart' Beyond the Brain: Ptmind Launches Beta of Kocoro, a Mac-Native AI Partner with Memory

Published Jun 3, 2026 11:00 AM ・ Updated Jun 14, 2026 4:48 AM ・ 9 min read ・ Source: PR TIMES

Ptmind launched the beta for Kocoro, a local AI partner for Mac that autonomously executes tasks and maintains 'proprietary memory' of a user's workflow, eliminating the need to repeatedly explain context.

Ptmind (Headquarters: Shinjuku-ku, Tokyo; CEO: Robin Zheng) announced the launch of the beta version of Kocoro, a local AI partner running on Mac, effective June 3, 2026.

Kocoro is an AI agent that autonomously executes tasks on Mac while acting as an 'AI partner' that 'remembers' the user's work context, allowing them to delegate daily operations without having to re-explain project backgrounds. It is characterized by keeping data on the Mac and releasing its core as open source.

Ptmind, which has honed customer understanding technology for over 15 years through the development of the Web experience optimization platform 'Ptengine,' is launching Kocoro as the next step to translate such understanding into 'decision and execution.'

Official Website: https://kocoro.ai/ja/ GitHub: https://github.com/Kocoro-lab/Kocoro

Key Highlights of this Release:

- Built on Ptengine Insights: This product leverages customer understanding technology refined over 15 years through the Web experience optimization platform 'Ptengine,' used by over 200,000 companies in 184 countries.

- Designed for Local Data Control: Understanding, files, and browser info are primarily kept on the Mac. The core is open-source, allowing users to verify operations, differing from many cloud-based AIs.

- 'Proprietary Memory' to Pick Up Where You Left Off: Because it remembers past interactions and decisions, users can resume work seamlessly.

- Free to Try: Includes a free tier. It uses a monthly fixed-fee model, eliminating unexpected additional charges.

Ptmind's AI Business: From 'Customer Understanding' to 'Decision and Execution'

Ptmind is expanding its AI business based on 'AI x Data' in two directions.

The first is utilizing customer understanding for management. Ptengine was born in 2011 as the world's first heatmap-based analytics tool for smartphone sites. Over 15 years, it has evolved into an all-in-one tool, accumulating web data practice globally, primarily in Japan and China. Understanding customer behavior, optimizing conversion, and integrating scattered data have all been areas covered by Ptengine.

The second is accelerating decision-making and execution using AI based on that understanding. This is what Kocoro, launching in beta on June 3, 2026, aims to do. It transforms knowledge and data into judgment power, turning routine tasks into time that can be focused elsewhere.

If Ptengine is a product that 'understands customers,' Kocoro is a product that 'acts based on that understanding.' Ptmind's consistent goal is to connect customer understanding to organizational competitiveness.

The Problem is 'Memory,' Not 'Intelligence'

Kocoro addresses the challenge of 'memory' rather than AI 'intelligence.' Even with advanced AI performance today, users often have to explain project backgrounds and previous decisions every time they use it.

While some approaches use 'longer context' to address this, Kocoro distinguishes itself by positioning this as 'Recitation' rather than memory.

Core Function: 'Proprietary Memory' – No Need for Repeated Explanations

Kocoro's core is 'Proprietary Memory.' It aims to eliminate the need to explain everything from scratch every time.

Mechanism: It accumulates daily events as a 'Understanding Graph.' While the user is away, Kocoro analyzes what happened that day (projects, people involved, decisions made, preferences, unfinished tasks) and leaves a connected 'Understanding Graph' on the user's Mac. Rather than saving raw conversation history, it stores only what is effective for future decisions, inspired by the 'episodic memory' of cognitive science.

Effect: Start from where you left off. When the user returns, Kocoro does not re-read history but responds based on grasp of tone, preferences, past decisions, team composition, and customer differences.

Value Kocoro Delivers: Delegate and Save Time

Beyond having memory, Kocoro provides the following functions as an AI agent operating on Mac:

- Call from Chat Apps (Slack / LINE / Lark): Even when away from the Mac, you can delegate tasks from your usual chat apps by messaging @Kocoro.

FACT BOX

Source: PR TIMES
Category: New Product
Products / services: Kocoro / Ptengine

Giving AI a 'Heart' Beyond the Brain: Ptmind Launches Beta of Kocoro, a Mac-Native AI Partner with Memory

⚡ Key Points

Key Highlights of this Release:

Ptmind's AI Business: From 'Customer Understanding' to 'Decision and Execution'

The Problem is 'Memory,' Not 'Intelligence'

Core Function: 'Proprietary Memory' – No Need for Repeated Explanations

Value Kocoro Delivers: Delegate and Save Time

FACT BOX

Editorial & Verification Standards

FAQ

Cite this article — HOW TO CITE

AI CRAWLER ACTIVITY