How I Built an AI Personal Assistant That Actually Works (And Saves Me $50K/Year)
Why I chose AI agents over hiring a PA, and how a simple n8n workflow eliminated my biggest productivity bottleneck
I have a confession: I absolutely suck at getting back to people.
Between 7+ calls per day, multiple Slack workspaces, an overflowing inbox, and what I'm pretty sure is the world's worst undiagnosed ADHD, important messages were constantly slipping through the cracks. Follow-ups weren't happening. Opportunities were dying in my inbox.
Sound familiar?
Instead of hiring a $50K personal assistant, I decided to build one. Using n8n and a few well-crafted AI agents, I created a system that now handles my entire communication workflow—and it's already made me 20-30% more efficient in just two weeks.
Here's exactly how I built it, what it does, and why every founder should consider something similar.
The Problem: Death by a Thousand Channels
Let me paint the picture. Every morning I'd wake up to:
50+ unread emails (mix of important deals and marketing spam)
15+ Slack DMs across multiple workspaces
Zero memory of which meetings from yesterday needed follow-ups
A growing sense of dread about what I was forgetting
The traditional solution? Hire a PA. But here's the thing—I didn't need someone to book my flights. I needed someone who could read my mind, understand context, and proactively surface what actually mattered.
That's when it hit me: AI agents could do this better than any human assistant ever could.
The Solution: A Multi-Agent AI System That Never Sleeps
I built my AI assistant using n8n (my go-to automation platform) with two main workflows that work together:
Workflow 1: The Email Manager
The Problem: My inbox was chaos. Important client emails buried under marketing noise.
The Solution: An AI agent that runs every time I get a new email and automatically categorizes it:
To Respond: Urgent emails requiring my attention
FYI: Important but not urgent
Marketing: Cold emails and newsletters
Notifications: Automated system emails
But here's the key—it doesn't just look at the email content. It checks my email history to see if I've ever corresponded with this person before. Context is everything.
The Result: My inbox became instantly manageable. No more decision fatigue about what to read first.
Workflow 2: The Personal Assistant Agent
This is where the magic happens. Every morning at 8am, this system:
Email Triage: Scans unread emails in my "To Respond" and "FYI" folders, prioritizes them by urgency, and flags anything related to active sales leads or partnerships.
Meeting Follow-up Check: Pulls all meetings from the last 3 days, checks the Fireflies transcripts to understand what happened, then verifies if I actually sent follow-ups. If not, it creates a to-do.
Slack Monitoring: Scans all my Slack mentions and DMs to find messages I haven't responded to, particularly in threads where follow-up is expected.
Master Orchestration: Takes all the above data, cross-references with previous to-dos (stored in Google Sheets), assigns priority levels, and sends me one consolidated Slack message with everything I need to action that day.
The Technical Stack (Simpler Than You Think)
n8n: Workflow automation platform
OpenAI API: For the AI agents (using cheaper models—this costs me $30-40/month total)
Fireflies: Call recording and transcription
Google Sheets: To-do tracking and historical context
Slack: Daily summary delivery
Gmail API: Email management
The beauty is in the breakdown. Instead of one massive, expensive workflow, I built smaller, focused agents that each do one thing well. This keeps costs low and makes the system much more reliable.
Real Results: From Chaos to Control
Since implementing this system two weeks ago:
Zero missed follow-ups from important meetings
30% reduction in time spent managing communications
100% capture rate of important action items
Significantly less anxiety about what I might be forgetting
Here's what my daily summary looks like:
Hi Max,
**PRIORITY EMAIL FOLLOW-UP REQUIRED:**
• Trigify partnership discussion with [Company] - requires technical follow-up
• Invoice follow-up needed for Q4 consulting work
**MEETING FOLLOW-UP PENDING:**
• Yesterday's investor call - send deck updates
• Demo with potential enterprise client - scheduling issue needs resolution
**SLACK MENTIONS REQUIRING RESPONSE:**
• Team discussion about new feature priority
• Customer success question in #support-escalations
Every item includes an estimated time to complete and urgency level. No more guessing what's important.
The Key Insights That Made This Work
1. Context Is Everything
The difference between a good AI agent and a great one is context. My system doesn't just look at individual emails—it checks my history with that person, cross-references meeting transcripts, and considers previous to-dos. This context is what makes the prioritization actually useful.
2. Break It Down Into Micro-Agents
Instead of building one complex workflow, I created specialized agents for each channel (email, Slack, meetings). This makes the system more reliable, easier to debug, and surprisingly cost-effective.
3. Human-in-the-Loop Is Critical
The system never sends emails or takes actions without my approval. It surfaces information and creates to-dos, but I maintain full control. This builds trust and prevents AI mistakes from damaging relationships.
4. Start With Your Biggest Pain Point
I started with email categorization because that was my biggest bottleneck. Once that worked, I expanded to meetings and Slack. Don't try to boil the ocean—pick one channel and perfect it first.
Where This Is Heading: The Future of AI Assistants
I'm already planning v2 features that will push this even further:
Calendar Integration: The system will automatically block time in my calendar to complete each to-do, turning suggestions into scheduled work.
Smart Email Drafting: Generate reply drafts for my approval, turning my daily summary into a one-click response system.
Proactive Relationship Management: Track relationship health across all channels and suggest touchpoints before they become urgent.
Cross-Platform Intelligence: Integrate with our CRM, Linear, and other tools to provide even richer context about what matters most.
The Bottom Line: This Changes Everything
We're at an inflection point where AI can genuinely augment human productivity—not just automate simple tasks, but actually think alongside us. This assistant doesn't just save me time; it makes me a better leader by ensuring nothing falls through the cracks.
The cost? About $40/month. The alternative? A $50K/year assistant who works banker's hours and can't read my meeting transcripts.
The real question isn't whether you can afford to build something like this. It's whether you can afford not to.
Want the technical details? I'm sharing the complete n8n workflows, prompts, and setup instructions with my newsletter subscribers. Everything you need to build your own AI assistant is in next week's deep-dive issue.
Check out a webinar I did on it →
Ready to level up your productivity? This is just the beginning. The teams that master AI-human collaboration first will have an unfair advantage in 2025.
What's your biggest communication bottleneck? Reply and let me know—I might feature solutions in future newsletters.
Very cool!
Oi on de está o json. para teste?