AI Agent · Project Management

Make PM a Little Easier

Drawn from real enterprise work. Internal systems and business data stay private, so this demo reconstructs the product design and the human–agent workflow with simulated data.

Create IssuePremortem Premortem AnalysisSprint Report

01 · Why this exists

Why we built this

  1. 01

    Inconsistent issue quality

    Different writing styles lead to extra clarification and rework.

  2. 02

    Risks discovered too late

    Potential blockers often surface only after execution begins.

  3. 03

    Manual sprint reporting

    Gathering and compiling the numbers takes time, and important details slip through.

02 · How it works

How AI fits into the workflow

Project management agent board screenshot

03 · Tool selection

Tool selection

Multi-turn dialogue

The draft needs iterative refinement.

Visual interface

Complex branches stay clear in the GUI.

04 · Why it works

Why it's designed this way

  1. 01

    Fit Existing Workflows

    Stay inside the team's current way of working. Don't invent a parallel process for AI.

  2. 02

    Keep It Simple

    If a conversation or a clear choice will do, don't add another step.

  3. 03

    Human-in-the-loop

    AI drafts and suggests. People still make the calls that matter.

05 · Choose how to explore

How to try it

Demo mode

No API key

Premortem analysis uses preset workflow for demonstration only; other features are powered by Dify.

Live mode

API key required

Add your own API key to explore all features live.

06 · Behind the demo

From enterprise project to public demo

Real Project

  • Enterprise Jira
  • Business data
  • Real team workflow
  • User feedback & iteration
↓↓

Public Demo

  • Jira-like interface
  • Simulated data
  • Dify workflows + Premortem demo/live
  • Same product workflow
© Shuyi · 2026From Plato to GPT.