Grab me. Throw things. Swap my hat.
Current casetoodo

Mingjian Shao

AI Product Manager · Enterprise AI Deployment (FDE)

I help teams put AI to work, from model access to working products.

Current case

Since 2013

Commercial, product, and digital-delivery experience

Enterprise AI

Learning-note products, evaluation, and operations

Open source

Agent tools with tests, releases, and documentation

About

I help small and mid-sized companies make AI useful in everyday work.

25/25

Competition projects coached — all submitted.

A demo before a build

I build a working demo to check what you need, then engineer it for your scale or work with your developers.

AI access for your team

I set up shared model access, accounts, and permissions so your team can use the best models at a sensible cost.My desktop AI workspace has one sign-in and central access control; it is in a controlled pilot.

AI quality you can check

I define generation rules, regression sets, and acceptance criteria so you can compare model quality and cost.I evaluated AI-generated learning notes and investigated generation failures and model costs.

Connected workflows and data

I turn workflows into Skills and connect systems through MCP tools with API docs, so people and agents use the same data.I built a Feishu–ERP toolchain for document sync and cross-system queries.

What I take on

Diagnostic consulting, building a project together, and team workshops.

Services

Things I’ve built

From publishing to enterprise AI

  1. Digital publishing from zero

    I started a digital publishing line from scratch, later developing ebooks, distribution channels, and a reading app.

  2. Phone-based reading

    I led a phone-based point-reading product with a paper stand and was first-named designer on its design patent, granted in 2021.

    Design patent
  3. Processes and CRM

    I built an eight-module CRM on Feishu, connecting customers, opportunities, contracts, and payments.

  4. Systems connected through MCP

    I began connecting Feishu and ERP data through MCP tools for document sync, meeting archives, and cross-system queries.

  5. Open tools for agent work

    I maintain another, loop, and OpenCode Annotate for coding-agent sessions, product workflows, and review feedback.

  6. A desktop AI workspace

    I built a desktop workspace that puts models, business tools, and Skills behind one sign-in with central access control; it is in a controlled pilot.

  7. A national competition demo

    I delivered application updates and the final demo as the sole technical member of a team that placed in the Top 10 of the Year, General Track, at the 2026 AI Pioneer National Competition hosted by Feishu.

I want information to flow through AI, with people keeping the core judgment, confirmation, and correction.

Latest notes

More projects

All projects
another

another

Every coding-agent session in one place: find, resume and organize them, @-mention across agents, or write one into another agent's native store to continue it instead of pasting a summary.

loop

loop

One explicit trigger, then artifact readiness chains deep-grill, deep-design and deep-build until the work reaches a confirmed spec or a verified result.

OpenCode Annotate terminal review in Dracula theme with two annotations and a reaction toolbar

OpenCode Annotate

An unofficial OpenCode plugin for annotating AI replies, documents, and webpages, with a bundled terminal reviewer and optional browser companion.