Zing Forum

Reading

PilotFlow: AI Project Operator in Feishu Group Chats, Trigger a Complete Workflow with One Sentence

PilotFlow is a Feishu plugin based on the Hermes Agent runtime. It automatically extracts project elements via natural language instructions and calls Feishu APIs to create real documents, tasks, and project entries, enabling seamless transition from group chat discussions to formal project deliverables.

飞书AI Agent项目管理协作工具自动化工作流LLM即时通讯生产力工具企业应用
Published 2026-05-02 20:45Recent activity 2026-05-02 20:50Estimated read 7 min
PilotFlow: AI Project Operator in Feishu Group Chats, Trigger a Complete Workflow with One Sentence
1

Section 01

PilotFlow: AI Project Operator in Feishu Group Chats, Trigger a Complete Workflow with One Sentence

PilotFlow is a Feishu plugin based on the Hermes Agent runtime. It automatically extracts project elements via natural language instructions and calls Feishu APIs to create real documents, tasks, and project entries. It addresses the pain points of scattered information in group chat collaboration and time-consuming manual project creation, enabling seamless transition from group chat discussions to formal project deliverables.

2

Section 02

Pain Points of Group Chat Collaboration and the Background of PilotFlow's Birth

When modern teams collaborate using instant messaging tools like Feishu, they face issues such as scattered information, loss of key content, and difficulty tracking follow-up actions. Starting a project requires manually creating spaces, organizing documents, and assigning tasks, which takes 30 minutes or even longer. PilotFlow was created to address this pain point by deeply integrating AI capabilities into Feishu group chats, allowing a single sentence to trigger a complete project creation workflow.

3

Section 03

Analysis of PilotFlow's Core Capabilities

PilotFlow is positioned as the "AI Project Operator in Feishu Groups", with core capabilities including:

  1. Intelligent Element Extraction: Understand user intent through large language models, automatically identify key information such as project goals, participating members, deliverable lists, and deadlines;
  2. Automated Deliverable Creation: Call Feishu APIs to generate real documents, tasks, and project entry messages;
  3. Intelligent Permission Management: Automatically open link access permissions for documents and @mention project members;
  4. Full Traceability: Detailed process logs record every step from instruction to deliverables.
4

Section 04

Technical Architecture Design of PilotFlow

PilotFlow adopts a plug-in architecture built on the Hermes Agent runtime:

  1. Underlying Runtime: Hermes provides an Agent environment, Feishu WebSocket gateway connection, and LLM scheduling capabilities;
  2. Core Plugin Layer: PilotFlow implements project management workflows and Feishu API integration, directly connecting to Feishu APIs via the lark_oapi SDK;
  3. Intelligent Reasoning Layer: Uses the mimo-v2.5-pro model (supports OpenAI-compatible interfaces and optimized for Chinese intent understanding) to identify needs and select tools for execution.
5

Section 05

Usage Flow and Experience of PilotFlow

Usage Flow:

  1. Natural Language Trigger: @PilotFlow in a Feishu group and describe project requirements;
  2. AI Understanding and Confirmation: Parse input to extract elements, and request supplementary information if necessary;
  3. Automated Deliverable Generation: Create formatted Feishu documents, member-associated tasks, and summary notification messages;
  4. Instant Collaboration Start: The entire process takes approximately 17 seconds, with efficiency significantly improved compared to traditional manual methods.
6

Section 06

Typical Application Scenarios of PilotFlow

Applicable to various work scenarios:

  1. Agile Development Initiation: Quickly create user story documents and development tasks;
  2. Cross-Department Collaboration Projects: Quickly establish project spaces, avoiding permission applications and tool switching;
  3. Meeting Minutes Conversion: Generate to-do tasks and assign them to relevant personnel;
  4. Brainstorming Implementation: Transform ideas into formal projects.
7

Section 07

Future Development Roadmap of PilotFlow

Development is divided into five phases: Phase1 (Completed): Basic plugin functionality; Phase2 (Completed): LLM-driven intent understanding and plan generation; Phase3 (Completed): Direct lark_oapi SDK connection, @mention, formatted documents, and automatic permission opening; Phase4 (In Progress): Confirmation gating, risk detection, and operation records; Phase5 (Planned): Multi-project spaces, Worker preview, and self-evolution capabilities.

8

Section 08

New Paradigm of AI-Enabled Collaboration: Value Summary of PilotFlow

PilotFlow redesigns the complete workflow from idea to execution, making AI a natural extension of team collaboration. It provides Feishu users with a convenient way to enjoy AI empowerment without switching tools or learning new systems. It heralds the future trend of deep integration between AI and collaboration tools, driving more intelligent and efficient collaboration.