How I Built an AI Personal Assistant That Works 24/7
Step-by-step beginner’s guide to building your own AI personal assistant in Telegram with n8n and OpenAI. Manage email, calendar, Notion, and more instantly — all from a single message.
It was 2:17 AM when the idea hit me.
I was lying in bed, supposed to be sleeping, when my brain decided it was the perfect time to remember: I forgot to email Alex about the project timeline. I also needed to schedule that call with the client for next week. Oh, and I had this brilliant idea for a Substack article that I definitely couldn't let slip away.
Sound familiar?
You know what happened next. I grabbed my phone, squinting at the bright screen, and started the familiar routine:
Open Gmail app → reread conversations to make sure nothing slips → check project in Drive → compose email → reread a few more times to be sure it’s clear → schedule send for tomorrow morning
Open Calendar app → create event → add details → set reminder for tomorrow
Open Notes app → jot down a half-coherent idea hoping I’ll understand it tomorrow → add a calendar reminder to revisit and draft while it’s still fresh
Twenty minutes of fumbling in the dark across multiple apps. And by the time I was done, I was wide awake.
But here's what really frustrated me: this wasn't a one-time thing. This was my life.
Whether I was stuck in traffic, at the market, walking in the park, or trying to enjoy a vacation, my brain would constantly interrupt with work thoughts. And each time, I'd have to navigate through multiple apps, deal with tiny keyboards, and waste my mental energy on administrative busywork.
The question became obvious:
What if I could just send a message and have all of this handled automatically?
Not hire an expensive human assistant. Not add another app to my workflow. But build something that could process everything through a single text or voice note.
So I did exactly that.
And after 5 days of obsessive testing, tweaking, and perfecting, I created an AI personal assistant that never sleeps and costs almost nothing to run.
What my AI Personal Assistant does
Instead of fumbling through apps, here’s what I get done by simply sending a message:
1. Calendar management:
Sets reminders instantly: "Remind me to call John tomorrow at 2 PM"
Creates detailed events: "Schedule lunch meeting with Sarah next Friday at 7 PM at that Italian place downtown"
Checks my availability: "What do I have scheduled Thursday afternoon?"
2. Email handling:
Reads my inbox: "What emails did I get last week that I didn't get to respond to?"
Sends emails: "Email Maya and tell her I'll be 15 minutes late to our meeting"
Drafts emails for review: "Draft an email to the client explaining our new pricing structure"
3. Information gathering:
Searches the web: "Find me the best sushi restaurants within 10 minutes of my home"
Gets news updates: "What's happening in AI this week?"
Research anything: "Find me statistics on remote work productivity"
4. Idea capture & processing:
This is the game-changer. When inspiration hits at random moments, I just send: "Idea for a new Substack series: what if I sent out a weekly email with a business idea you could build with AI and a step-by-step guide on how to do it?”
The assistant doesn't just save it, it:
Categorizes the idea
Clarifies and expands on it
If it's a creative task, drafts an outline or starting content
Stores everything organized in Notion
5. Document management:
This one solved a specific pain point for me. I get a lot of invoices, contracts, and random files that I need stored but never want to deal with manually. Now I just drop any document into the chat and it automatically:
Uploads to the right Google Drive folder
Renames it with current month/year + original name
Keeps everything organized without me thinking about it
No more "I'll file this later" and then forgetting where I put it.
How I built it (the technical breakdown)
Let me walk you through the pieces that make this whole thing run.
The tool stack:
N8N (workflow automation platform)
OpenAI API (the AI brain, but you can use any other LLM)
Telegram (the chat interface - though you can use Slack or WhatsApp)
Google Cloud (Calendar, Gmail, Drive, Sheets - or you can use Outlook + Airtable + other alternatives)
SerpAPI (for web searches)
Notion (used for idea storage and processing, but you can swap it for Google Sheets or some other tool if you prefer)
The architecture:
Two main n8n workflows power this system:
Main AI Assistant Workflow - Handles all real-time requests through Telegram
Idea Processing Workflow - Takes captured ideas and processes them in Notion
You’ll get both ready-made, so you can import them and have it set up in less than 2 minutes.
How it works:
When you send a message to your Telegram bot, here's what happens:
Message received → N8N catches it instantly (if it’s a document, it uploads to Drive; if it’s a voice note, it’s transcribed into text; if it’s already text, it just passes through as it is)
AI analysis → OpenAI determines what type of request this
Action routing → Based on the analysis, it routes to the appropriate service
Task execution → Performs the actual task (sending email, creating calendar event, etc.)
Confirmation → Sends you confirmation that the task is complete
The magic happens in the mega-prompt I created that powers the AI agent. This prompt gives the AI the context and capabilities to understand your requests and determine exactly what actions to take.
What it actually costs to run this system
N8N subscription: $20/month (or free if you self-host and are comfortable with the technical setup)
OpenAI credits: ~$0.0088 per task request (yesterday, I tested this system with over 50 requests and spent $0.44 in AI credits)
Total monthly cost projection:
10 tasks per day = ~$2.64/month in AI credits for 300 tasks
Total monthly cost: ~$23 (that covers n8n too, which you can also use for many other automations to save time)
Compare this to hiring a human assistant at even $15/hour for just 2 hours per week, and you're looking at $120/month minimum.
But the real value isn't the money saved. It's the time and mental energy recovered.
The real cost of manual task management
Before I hand you everything you need to set this up for yourself, it’s worth pausing to look at what these tiny tasks actually take out of us.
Creating a calendar event manually: 1-2 minutes per task
Writing and sending an email: 5-10 minutes per email
Searching for information: 5-10 minutes per search
Capturing and organizing ideas: 2-4 minutes per idea
Managing documents: 2-3 minutes per document
Even at just 10 tasks a day (and most of us do far more) that’s half an hour lost to administrative overhead. Over time, it adds up to 3–4 hours a week, or more than 180 hours a year.
But time isn't the only cost. On top of it, you carry the mental overhead of keeping track of unfinished tasks, the constant context switching of jumping between apps, the opportunity cost of ideas that slip away, and the stress of feeling like you’re forgetting something.
That’s the real tax of manual busywork.
Why this changes how I work
It stripped out the small, repetitive tasks that used to clutter my day. Now I just send one message and the assistant handles the rest:
Instead of managing multiple apps and interfaces, I have one conversation with an AI that handles everything in the background.
Instead of losing ideas because I couldn't write them down fast enough, every thought gets captured and processed instantly.
Instead of spending mental energy on administrative tasks, I focus on work that moves the needle.
The results in my first week:
~4 hours of admin time saved
24 ideas captured that could have been lost
No missed follow-up emails
Noticeably less mental overwhelm
The system just runs in the background, handling things I used to have to remember to do.
What you get in the implementation package
✅ The battle-tested automation workflows: Two complete N8N workflow JSON files - the exact system I built for myself, tested across 5 days with 150+ edge cases - ready for you to import in seconds.
✅ Step-by-step tool connection guides: Most people just drop the automation and leave you to figure out the setup. Here, you get the complete connection steps so everything works smoothly. (And don’t worry if you’ve never dealt with APIs, tokens, or service accounts before. I’ll guide you through every click, with screenshots and videos.)
✅ Step-by-step implementation walkthrough: A clear, start-to-finish guide of every step you need to set up the system, so you can get it running without guesswork.
✅ The master AI Agent mega-prompt: The exact prompt that powers the assistant and keeps the whole system running smoothly, with room for you to add your own context and adapt it to your needs.
All of this is what I wanted when I built it for myself, and now you can have the same setup running in less than 60 minutes, including the time it takes to connect each tool.
Let me show you how.
Complete implementation guide
This is a beginner-friendly, step-by-step walkthrough. You won’t be left guessing what to do or where to click. Every step is spelled out clearly, with the exact buttons, fields, and settings you need to use, so you can follow along even if you’ve never set up an API or workflow before.
Step 1: Account setup
Create your N8N account
Go to n8n.io and sign up for the 14-day free trial
Choose the cloud version (unless you're comfortable self-hosting)
Once logged in, click "Create Workflow"
Import my workflows
Click the three dots in the top right corner
Select "Import from File"
Download the two JSON files I’m providing.
Import each one into n8n: