Building an AI Personal PM Manager using Make.com AI Agent for Task Management via Telegram — Advice Needed

How To
,
1

Hello Make.com community!

I am working on creating an AI-powered personal project manager assistant.

Main goal:
Seamless communication and task management through Telegram.

What the assistant should be able to do right now:

  1. Create tasks from messages I send via Telegram.

  2. Provide me information about tasks and their details (e.g., whether the task exists, what it contains, tasks related to a specific client, etc.).

  3. Update tasks — including status, assignees, task details, and more.

Here is a description of the system as ChatGPT presented it:

Tasks Management

  • Task Creation

    • Accepts input data: task name, client, project, assignee, deadline, status, comment, tag.
    • Automatically creates a new task in Data Store with a unique ID.
    • Correctly links the task to the client and project via their IDs.

  • Duplicate Checking

    • Checks before creation if the same task already exists to avoid duplicates.

  • Task Updates

    • Allows changing deadline, status, assignee, description, client, project, tag, and other attributes.
    • If an update is ambiguous, the bot asks for clarifications.
    • Stores the date of the last modification (task_modified).

  • Task Display

    • Can output a filtered list of tasks by client, project, status, deadline.
    • Supports search and sorting.

Current challenge:

Previously, ChatGPT stored this info internally, so fetching task details was fast. Now, getting necessary task info takes longer because the process is slower.

Current logic:

  • There is a main agent receiving Telegram messages.
  • A Data Store holds entities for clients, projects, and tasks.
  1. Task Statuses:
  • To-Do — task planned, needs execution
  • In Progress — task in progress
  • In Review — task under review
  • Backlog — postponed tasks
  • Sent to Client — task sent to the client
  • Waiting for Client — waiting for client response/info
  • Revisions — task under revision
  • Done — completed
  • Archived — archived, no longer active
  • On Hold — paused
  1. Main Entities (Data Store tables):

Clients:

  • id (uuid or key)
  • name (client name)
  • created_at (creation date)

Projects:

  • id (uuid or key)
  • name (project name)
  • client_id (linked client)
  • client_name (for convenience)
  • created_at (creation date)

Tasks:

  • id (uuid or key)
  • task_name
  • task_deadline
  • task_client (id or key)
  • task_client_name
  • task_project (id or key)
  • task_project_name
  • task_status (from the statuses above)
  • task_comment
  • task_assigned
  • task_tag (e.g., “Design”)
  • task_resultlink (link to results or documents)
  • task_modified (last update date)
  • created_at (creation date)

Please advise how to implement the following:

  1. Use one main agent and multiple sub-scenarios.
  2. Have the ability to communicate with ChatGPT via Telegram just as directly, without worrying about where and how all the data is stored.
2

Take a look at FamilyMate.AI - Make AI Agent, which uses Telegram and multiple sub-scenarios.

1 Like