Tier 1 learning layer

Work Learn

Learn the English your work already needs

Work Learn turns the English you already use with AI into a personal course. It captures reusable expressions from real agent conversations through an MCP server, a skill, the CLI or the macOS Companion, then rebuilds them as practice around your own work.

LiveLearning layer for AI workflows· 2026-08-16
AIMCPAstroTypeScriptTailwind CSS

See the proposed context handoff to zeus.

Problem

Generic English practice does not stick

Most English learning is detached from the work you actually do. The phrases you need in technical discussions show up in AI conversations, then disappear. A notebook captures words, not the next time you can use them.

Approach

Capture the good sentence, then reuse it

Work Learn captures real AI work conversations and selects expressions worth reusing, not every word. It explains the natural version, shows an example shaped around your codebase, and brings the expression back in multiple practice formats when you need it.

What it does

01

Multi-channel capture

An MCP server, a skill, a terminal CLI and a macOS Companion all feed the same learning layer.

02

Expression selection

The layer keeps reusable phrases from real conversations, not a raw transcript or a word list.

03

Work-shaped examples

Explanations and examples are rebuilt around your code, your API layers and your queue.

04

Practice that returns

Saved expressions resurface as multiple-choice, fill-in and scenario prompts when they are useful again.

What is real today

Taken from the current catalog description and the public landing copy.

Capture
MCP, skill, CLI, Companion
Source
real AI work conversations
Selection
reusable expressions
Practice
choice, fill-in, scenario

Boundary

It is not a notebook dump, a generic dictionary or a fixed curriculum. It builds a personal corpus from your own work, and the proposed zeus DataDomain handoff is a direction record, not an implemented integration.

Directional link

This product →

feeds · Proposed

Zeus

Conversation context into DataDomain

Evidence (1)
  • docs/tier0/Tier0-个人上下文层与PersonalAgent-讨论备忘-2026-11-19.md

    Direction memo: Work Learn as a producer, Zeus DataDomain as the single context source; not implemented.

See the proposed context handoff to zeus.

Open the product map