🚀 Feature Proposal: Personal Assistant Agent with Calendar, Email, and Notes Integration #83

New Issue

GiteaMirror · 2025-11-06T14:32:35-06:00

GiteaMirror commented

2025-11-06 14:32:35 -06:00

Originally created by @RajalakshmiMadhavan on GitHub (Sep 12, 2025).

Description:
I’d like to propose a new app for the Personal Assistant category: Calendar, Email & Notes Assistant.

Idea:
✅ User connects their calendar, email, and notes accounts via OAuth.
✅ The LLM helps users manage their tasks by understanding natural language commands either by text or voice message such as:
“Schedule a meeting with John next Tuesday at 3 PM.”
“Find unread emails from my manager.”
“Create a note about project ideas for Q4.”
✅ The assistant can:
🗓 Schedule, view, and update events.
📧 Draft, send, and summarize emails.
📝 Create, edit, and retrieve notes based on context and queries.

Why this is useful:
Helps users manage daily productivity tasks without toggling between apps.
Saves time by automating scheduling, reminders, and note-taking.
Improves organization by providing context-aware suggestions.
Enhances user experience by offering a conversational interface for task management.

Tech Stack (proposed):
Backend: Python + LLM (OpenAI API / Ollama for open-source alternatives)
Integrations:
Google Calendar API
Gmail API
Notes API (e.g., Evernote, Notion, or local file system)
Frontend: Streamlit or Flask for user interactions and simple UI
Authentication: OAuth 2.0 for secure data access
Output: Markdown or downloadable summaries of tasks, emails, and notes

Next Steps:
If maintainers approve, I can create a PR with:
app.py for the main application logic and UI
requirements.txt for dependencies
README.md with setup instructions, usage examples, and security notes

Originally created by @RajalakshmiMadhavan on GitHub (Sep 12, 2025). **Description:** I’d like to propose a new app for the Personal Assistant category: Calendar, Email & Notes Assistant. **Idea:** ✅ User connects their calendar, email, and notes accounts via OAuth. ✅ The LLM helps users manage their tasks by understanding natural language commands either **by text or voice message** such as: “Schedule a meeting with John next Tuesday at 3 PM.” “Find unread emails from my manager.” “Create a note about project ideas for Q4.” ✅ The assistant can: 🗓 Schedule, view, and update events. 📧 Draft, send, and summarize emails. 📝 Create, edit, and retrieve notes based on context and queries. **Why this is useful:** Helps users manage daily productivity tasks without toggling between apps. Saves time by automating scheduling, reminders, and note-taking. Improves organization by providing context-aware suggestions. Enhances user experience by offering a conversational interface for task management. **Tech Stack (proposed):** **Backend:** Python + LLM (OpenAI API / Ollama for open-source alternatives) **Integrations:** Google Calendar API Gmail API Notes API (e.g., Evernote, Notion, or local file system) **Frontend:** Streamlit or Flask for user interactions and simple UI **Authentication:** OAuth 2.0 for secure data access **Output:** Markdown or downloadable summaries of tasks, emails, and notes **Next Steps:** If maintainers approve, I can create a PR with: app.py for the main application logic and UI requirements.txt for dependencies README.md with setup instructions, usage examples, and security notes

GiteaMirror closed this issue

2025-11-06 14:32:35 -06:00

GiteaMirror commented

2025-11-06 14:32:36 -06:00

@Shubhamsaboo commented on GitHub (Sep 12, 2025):

Thanks for proposing the idea. It looks good to me.

This would be my preference for the tech stack:

OpenAI API
Streamlit frontend

rest follow the structure you suggested:

app.py for the main application logic and UI
requirements.txt for dependencies
README.md with setup instructions, usage examples, and security notes

@Shubhamsaboo commented on GitHub (Sep 12, 2025): Thanks for proposing the idea. It looks good to me. This would be my preference for the tech stack: - OpenAI API - Streamlit frontend rest follow the structure you suggested: - app.py for the main application logic and UI - requirements.txt for dependencies - README.md with setup instructions, usage examples, and security notes

GiteaMirror commented

2025-11-06 14:32:36 -06:00

@RajalakshmiMadhavan commented on GitHub (Sep 15, 2025):

Hi @Shubhamsaboo ,
Thank you for your response! I’m confident I’ll be able to complete this contribution within the next two weeks. I have some other work at the moment, but I’ll manage both and ensure timely progress.
Looking forward to contributing!

Will check your preference tech stack and come back incase of any queries on this.

@RajalakshmiMadhavan commented on GitHub (Sep 15, 2025): Hi @Shubhamsaboo , Thank you for your response! I’m confident I’ll be able to complete this contribution within the next two weeks. I have some other work at the moment, but I’ll manage both and ensure timely progress. Looking forward to contributing! Will check your preference tech stack and come back incase of any queries on this.

GiteaMirror commented

2025-11-06 14:32:36 -06:00

@Shubhamsaboo commented on GitHub (Sep 15, 2025):

Sounds good. Looking forward to it!

@Shubhamsaboo commented on GitHub (Sep 15, 2025): Sounds good. Looking forward to it!

GiteaMirror commented

2025-11-06 14:32:36 -06:00

@RajalakshmiMadhavan commented on GitHub (Sep 16, 2025):

@Shubhamsaboo I am not able to test this in local since my credits for open AI has been exhausted. Can I go ahead and use Ollama ?

@RajalakshmiMadhavan commented on GitHub (Sep 16, 2025): @Shubhamsaboo I am not able to test this in local since my credits for open AI has been exhausted. Can I go ahead and use Ollama ?

GiteaMirror commented

2025-11-06 14:32:37 -06:00

@RajalakshmiMadhavan commented on GitHub (Sep 18, 2025):

@Shubhamsaboo Have few concerns in this implementation:

Firstly, can't use open ai or other models for this since it can directly access our gmail/calendars which is a high risk in terms of security
Can't use ollama also since it can't communicate with external API services like gmail/calendar. After giving prompt to ollama, we might need to hit external api manually for things to be done which is defeating the purpose of using AI models.

Need your help on how to take this forward. Both giving access to AI or revoking the access(defeating the purpose of it) is the concern here. Your thoughts ?

@RajalakshmiMadhavan commented on GitHub (Sep 18, 2025): @Shubhamsaboo Have few concerns in this implementation: 1. Firstly, can't use open ai or other models for this since it can directly access our gmail/calendars which is a high risk in terms of security 2. Can't use ollama also since it can't communicate with external API services like gmail/calendar. After giving prompt to ollama, we might need to hit external api manually for things to be done which is defeating the purpose of using AI models. Need your help on how to take this forward. Both giving access to AI or revoking the access(defeating the purpose of it) is the concern here. Your thoughts ?