I used to spend 12+ hours a week on meeting-related tasks. Taking notes, writing summaries, sending follow-ups, updating project trackers — the whole circus. In January 2026, I decided enough was enough and built an AI-powered meeting workflow that handles almost everything automatically.
Here's exactly how I did it, step by step.
The Problem: Death by Meetings
Let me paint the picture. As a product lead at a mid-size SaaS company, my typical week looked like this:
- 15-20 meetings across different time zones
- 3-4 hours spent writing meeting notes after the fact
- 2 hours drafting follow-up emails
- 1-2 hours updating Notion, Jira, and Slack with action items
That's roughly 25% of my work week spent on administrative overhead from meetings — not the meetings themselves. I was essentially a very expensive secretary.
Something had to change.
Step 1: AI Meeting Recording and Transcription
The first piece of the puzzle was getting accurate, automatic transcriptions of every meeting. I'd tried basic transcription tools before, but they always felt clunky — missing context, butchering technical terms, and requiring heavy editing afterward.
That changed when I started using Fireflies.ai.
What Made Fireflies Different
Fireflies joins your meetings automatically (Google Meet, Zoom, Teams — it doesn't matter) and creates transcriptions that are genuinely usable. But the transcription is just the starting point. Here's what actually saves me time:
AI-Generated Summaries: After every meeting, Fireflies produces a structured summary with key topics discussed, decisions made, and action items identified. I used to spend 20-30 minutes writing these manually. Now they're ready before I've even closed the Zoom window.
Smart Search Across Meetings: Last week, a colleague asked me "What did we decide about the pricing tier restructure?" Instead of digging through my notes, I searched across all my meeting transcripts and found the exact moment we made that decision — with full context.
Action Item Tracking: Fireflies automatically identifies who committed to doing what. This alone has reduced our "I thought you were handling that" moments by probably 80%.
CRM Integration: For sales calls, meeting notes automatically sync to our CRM. Our sales team went from spending 45 minutes per call on admin to about 5 minutes of review.
My Fireflies Setup
Here's my actual configuration:
- Auto-join: Enabled for all calendar events with a video link
- Summary style: Set to "detailed" for client calls, "concise" for internal standups
- Notifications: Slack webhook that posts summaries to relevant channels
- Privacy: I always inform participants that AI is taking notes (this is important — more on ethics later)
The cost? About $18/month on the Pro plan. Considering I was spending 3-4 hours a week on manual notes (at my hourly rate, that's... a lot more than $18), the ROI was immediate.
Step 2: Voice-First Documentation
Here's where things got interesting. Even with automated meeting notes, I still had a documentation problem. Between meetings, I'd have ideas, decisions, and context that needed to be captured. But sitting down to type everything out felt like yet another chore.
Enter Typeless — an AI voice dictation tool that fundamentally changed how I create documentation.
Why Typeless Clicked for Me
I'd tried voice-to-text before (Dragon, Google's built-in dictation, etc.), and it always felt like I was fighting the tool. Typeless is different because it understands context.
Here's what I mean: I can say something like "Update the sprint retro notes — the team agreed to move to two-week cycles starting March, and Sarah is owning the migration plan" and Typeless doesn't just transcribe those words. It formats them properly, understands the structure, and produces clean, readable text.
My daily workflow with Typeless:
- Morning brain dump (5 minutes): I dictate my priorities, blockers, and any overnight thoughts while making coffee
- Post-meeting capture (2-3 minutes per meeting): Immediately after a call, I dictate additional context that the AI summary might have missed — the political dynamics, my gut feelings, the subtext
- End-of-day review (5 minutes): I dictate a quick summary of what got done and what's carrying over
The numbers: I type at about 70 WPM. I speak at about 150 WPM. That's more than 2x faster, and because Typeless handles formatting and context, the output requires minimal editing. I estimate this saves me 45-60 minutes per day.
Combining Fireflies + Typeless
This is where the magic happens. Here's my actual workflow for a typical meeting:
- Before the meeting: I dictate prep notes into Typeless (talking points, questions, context I need to remember)
- During the meeting: Fireflies records and transcribes everything
- After the meeting (within 5 minutes): I dictate my personal takeaways into Typeless — things like "The client seemed hesitant about the timeline, need to follow up with a revised proposal" or "John's suggestion about the API redesign is worth exploring, create a spike ticket"
- Automated: Fireflies summary + my dictated notes get combined in Notion via a simple Zapier workflow
The result? Complete, nuanced meeting documentation that takes me about 5 minutes of active effort per meeting, compared to the 20-30 minutes I used to spend.
Step 3: Automating Follow-Ups
With transcriptions and notes handled, the next bottleneck was follow-up communications. After every meeting, someone needs to send a summary email, update the project tracker, and notify relevant stakeholders.
I built a simple automation using Fireflies' webhook + Zapier:
- Meeting ends → Fireflies generates summary and action items
- Webhook fires → Zapier receives the structured data
- Draft email created → Using the summary, a follow-up email draft appears in my inbox for review
- Notion updated → Action items are added to the relevant project board
- Slack notification → Team channel gets a summary with @mentions for action item owners
I still review and send the emails manually (I'm not ready to fully automate external communications), but the drafts are good enough that I usually just hit send with minor tweaks.
Step 4: The Weekly Review
Every Friday, I do a 15-minute weekly review that used to take over an hour:
- Pull up Fireflies dashboard → See all meetings from the week, total time spent, key decisions
- Dictate weekly summary into Typeless → Speaking through what happened is faster than writing, and it helps me process the week
- Review action items → Fireflies' tracker shows me what's been completed and what's overdue
- Plan next week → Dictate priorities and meeting prep for Monday
The Results: By the Numbers
After 6 weeks of running this system:
| Metric | Before | After | Savings |
|---|---|---|---|
| Weekly meeting admin time | 12 hours | 3 hours | 9 hours |
| Notes quality (self-rated) | 6/10 | 9/10 | — |
| Missed action items/week | 3-4 | 0-1 | ~85% reduction |
| Follow-up email time | 15 min/meeting | 3 min/meeting | 80% reduction |
| Documentation completeness | ~60% | ~95% | — |
That's 9 hours per week returned to actual productive work. Over a year, that's roughly 468 hours — or about 12 full work weeks.
Lessons Learned
1. Always Disclose AI Recording
I make it a point to inform every meeting participant that AI is taking notes. Most people appreciate it (they get the notes too!), and it's the right thing to do. Some organizations have policies about this, so check first.
2. Don't Over-Automate
I deliberately kept humans in the loop for external communications. AI-generated follow-ups are great drafts, but they sometimes miss tone or political nuance. Review before sending.
3. The Voice Dictation Habit Takes Time
It took me about two weeks to get comfortable dictating instead of typing. The first few days felt awkward, like talking to myself. Now it's second nature, and I can't imagine going back.
4. Start Small
Don't try to automate everything at once. I started with just Fireflies for transcription, got comfortable with that, then added Typeless, then built the automations. Each step took about a week to integrate into my routine.
Cost Breakdown
- Fireflies.ai Pro: ~$18/month
- Typeless: ~$12/month
- Zapier (for automations): ~$20/month (I was already paying for this)
Total: ~$50/month for 9 hours/week of time savings. That's roughly $1.28/hour saved. I'll take that trade any day.
What's Next
I'm currently experimenting with:
- Using Fireflies' API to build custom integrations with our internal tools
- Training Typeless on our company's specific terminology for even better accuracy
- Building a meeting analytics dashboard to identify which recurring meetings are actually valuable
The meeting workflow was my first major AI automation project, and it's given me confidence to tackle other time-consuming processes. If you're drowning in meeting overhead, I'd strongly recommend starting with transcription automation — it's the highest-ROI first step.
What does your meeting workflow look like? Have you automated any parts of it? I'd love to hear what's working for you in the comments.
📬 Want more practical AI workflow tips? I write a weekly newsletter covering the best AI tools and automation strategies for professionals. Subscribe to AI Product Weekly — no fluff, just actionable insights.
Disclosure: Some links in this article are affiliate links. If you purchase through them, I may earn a small commission at no extra cost to you. I only recommend tools I personally use and believe in.
Top comments (0)