Solo UnicornMy AI Meeting System — How One Person Handles 20 Meetings a Week
My AI Meeting System — How One Person Handles 20 Meetings a Week
Last week I counted the meetings on my calendar: 20, spread across Monday through Friday. Four client demos, three partnership discussions, two Solo Unicorn Club group sessions, four JewelFlow customer onboardings, five ArkTop AI project syncs, and two vendor negotiations.
When you're running three business lines solo, that's just the meeting density you get.
Back in early 2025, before I had any system for managing meetings, I was in at least 3 meetings a day. After each one, I'd manually take notes, compile action items, and send follow-up emails. The "post-meeting processing" alone ate 2–3 hours every day. By Friday I'd look at my to-do list and find several follow-ups from the previous week still undone — not because I forgot, but because there wasn't enough time.
I started building my AI meeting system in June 2025. Nine months later, the results: with 20 meetings per week, post-meeting processing dropped from 2–3 hours a day to 30–40 minutes, and the follow-up miss rate went from roughly 15% to near zero.
This article breaks down the entire system.
Background: Why Does a Solo Founder Need This Many Meetings?
Some people ask: why can't a solo business just have fewer meetings?
The answer is: the stage of business dictates it. ArkTop AI does B2B luxury retail, where maintaining client relationships depends entirely on person-to-person communication. JewelFlow is in a customer acquisition phase where every new client needs onboarding. The Solo Unicorn Club has regular group activities that require my involvement.
I've already cut everything I can. The remaining 20 per week are "drop-the-ball-if-you-skip-them" essential meetings.
Since the number can't come down, the only option is to find efficiency gains in how meetings are handled.
Core Approach: Three Principles
Principle 1: Let AI Handle Pre-Meeting Prep — Walk in with Context
An unprepared meeting is a waste of everyone's time. But with 20 meetings a week, you can't spend 30 minutes manually prepping for each one.
My approach: each morning I spend 10 minutes reviewing the day's meeting schedule, with Claude generating a briefing for each meeting.
The specific workflow:
- Google Calendar syncs the day's meetings to n8n automatically
- n8n triggers the Claude API, which does the following for each meeting:
- Pulls the last meeting summary for the contact/company (stored in Notion)
- Pulls key information from relevant email threads
- Generates a one-page briefing: what was discussed last time, outstanding action items, suggested agenda for this meeting
- All briefings are compiled into a single Slack message pushed to me
This way, I walk into every meeting with full context — no need to dig through chat logs and emails to remember "where did we leave off."
Principle 2: Fully Automate In-Meeting Recording — Just Be Present
During meetings, the only thing I do is focus on the conversation itself. Note-taking, recording, and summarizing key points are all handled by AI tools.
The tool selection was a journey. I started with Otter.ai Pro ($16.99/month), then switched to Fireflies.ai Pro ($10/month with annual billing).
The reason for switching was practical: Fireflies Pro is cheaper, and its CRM integration and auto-summary features better fit my needs. Otter does real-time captions better, but my meetings are primarily in English and I don't need live captions, making Fireflies the better value for my use case.
Fireflies does three things:
- Auto-recording and transcription: Joins Zoom/Google Meet calls, records the entire session, generates a full transcript
- AI summaries: Automatically generates meeting highlights, action items, and key decisions after the meeting
- Search: Full-text search across all meeting records from the past nine months — searching for "the meeting where we discussed pricing with Client A" returns instant results
I don't need to split my attention taking notes during meetings. My full focus stays on the conversation. This seemingly small change has a very noticeable impact on communication quality — especially in negotiations and client demos where high engagement is critical.
Principle 3: Auto-Trigger Post-Meeting Follow-Ups — Don't Rely on Memory
The moment a meeting ends is when balls get dropped most easily. You promised a client "I'll send the quote later," but the next meeting starts immediately, and by evening you've forgotten.
My automation flow:
- Fireflies sends generated action items to n8n via webhook
- n8n uses the Claude API to classify each action item:
- Do immediately (< 5 minutes, e.g., sending an email): Goes straight to today's to-do list
- Schedule for later (needs a time block, e.g., preparing a proposal): Automatically creates a Notion task with a suggested due date
- Waiting on others (need information from the other party): Automatically sends a follow-up reminder email, with a 3-day check-in scheduled
- Every day at 5 PM, a Slack summary is pushed: all action items generated from today's meetings + their status
The key design principle: the default action for follow-ups is "do it automatically," not "remind me to do it." Emails that can be auto-sent get auto-sent. Tasks that can be auto-created get auto-created. I only step in when judgment is needed.
Tool Stack Breakdown
| Use Case | Tool | Monthly Cost | Why I Chose It |
|---|---|---|---|
| AI meeting recording + transcription + summaries | Fireflies.ai Pro | $10 (annual) | Unlimited transcription, high-quality AI summaries, easy CRM integration |
| Pre-meeting briefing generation | Claude API (Sonnet) | ~$8 | Pulls historical context + generates briefings, 5–8 calls per day |
| Automation workflow | n8n (self-hosted) | ~$5 (server cost) | Connects Calendar, Fireflies, Notion, and email |
| Meeting record storage + task management | Notion | $0 | Structured storage for all meeting summaries and action items |
| Calendar management | Google Calendar (free) | $0 | Single entry point for all meetings |
| Team comms + notifications | Slack (free tier) | $0 | Receives briefings and daily summaries |
| Video conferencing | Zoom (Pro) | $13.33 (annual) | Used for most external meetings |
| Total | ~$36/month |
Including occasional Claude API usage spikes, the monthly average comes to $40–$45.
Why not Fathom? Fathom's free tier does solid recording and summaries, but its API/webhook capabilities are limited and don't integrate well with n8n for downstream automation. Fireflies has a more open API, making it better suited for building automated workflows.
Why not Otter Business? What Otter Business does for $20/month, Fireflies Pro does for $10/month. Otter's real-time collaboration features aren't useful to me (since it's just me).
Real Numbers
Nine months of operational data (June 2025 – February 2026):
Time Saved:
- Pre-meeting prep: From 15–20 minutes of manual research per meeting to 10 minutes per day reviewing AI briefings. At 20 meetings/week, that's ~4.5 hours saved weekly
- Post-meeting processing: From 2–3 hours/day to 30–40 minutes. ~8 hours saved weekly
- Total: 12+ hours saved per week
Quality Metrics:
- Follow-up miss rate: From ~15% to < 2% (occasionally the AI summary misses an offhand verbal commitment, but it's rare)
- Client feedback: One client proactively said, "Your follow-up turnaround is way faster than it used to be" (they don't know it's just me)
- Meeting search: 800+ meeting records accumulated over nine months, averaging 4–5 historical meeting searches per week
Which Meetings I Choose to Skip:
Not every meeting is worth attending. My criteria:
- Information-only meetings (someone sharing updates, industry webinars): Let the Fireflies bot attend on my behalf, then I read the summary. Saves 2–3 meetings per week
- Discussions that can be async: Convert to a Loom video or written response. Saves 1–2 meetings per week
- Meetings with no clear agenda: Politely decline, suggest they send an agenda first. This rule made roughly 30% of "let's just chat" meetings disappear on their own
In practice, out of the 20 meetings per week on my calendar, I personally attend about 15. The other 5 are handled by the AI bot or converted to async.
Lessons from Mistakes
Mistake 1: Fireflies bot joining meetings without advance notice.
One time during a client call, the Fireflies bot auto-joined the Zoom meeting. The client asked, "Who is this Fireflies Notetaker?" I wasn't prepared to explain, and it was awkward.
My fix: I now include a standard note in every Calendar invite — "This meeting uses AI-assisted note-taking; recordings are for internal reference only." I also renamed the bot in Fireflies settings to "Jesse's Notes Assistant," which people accept much more easily than the default name.
Mistake 2: AI summaries missed a critical verbal commitment.
In one meeting, a client casually said at the end, "Let's do a 10% discount on the quote." Fireflies' AI summary didn't flag this as an action item because it didn't look like a typical task. I ended up sending the quote at full price, and the client felt I wasn't reliable.
The lesson: I've developed a habit of verbally summarizing "the key points we agreed on today" at the end of every meeting. This both confirms mutual understanding and makes it easier for the AI to capture these points in the summary.
Mistake 3: Auto-generated follow-up emails sounded too robotic.
The initial Claude-generated follow-up emails used a cookie-cutter template tone. Some clients felt like they were communicating with a bot.
The fix: I spent an afternoon writing 5 email style templates (formal partnership, casual follow-up, thank-you, reminder, proposal submission). Claude automatically selects the appropriate template based on meeting type. I also require the email to reference one specific detail from the meeting, making the content feel more personal.
Mistake 4: Confidential information in stored meeting records.
Fireflies stores all recordings in the cloud by default. I once realized in hindsight that some recordings contained client pricing strategies and unreleased product information — what if Fireflies had a data breach?
Current approach: For sensitive meetings (involving pricing, legal matters, competitive strategy), I disable auto-recording. Instead, I record locally and process transcription and summaries via the Claude API, keeping data off third-party servers. About 3–4 meetings per month follow this pattern.
Advice for Getting Started
Step 1: Install an AI meeting assistant and use it for two weeks.
Fireflies' free tier includes 800 minutes of transcription per month — more than enough to try it out. Or Otter's free tier with 300 minutes per month. No automation setup needed — just experience what it feels like to not take your own notes.
Step 2: Solve the follow-up problem.
Use the simplest method: after each meeting, copy the AI-generated action items into your to-do list. Manual is fine — first build the habit of "processing action items immediately after every meeting."
Step 3: Build automation as needed.
When you find that manually copying action items has become a burden (usually around 10+ meetings per week), then consider setting up automation with n8n or Zapier. Don't build the full system upfront — automating before your needs are clear is just digging yourself a hole.
Final Thoughts
Meeting management for a solo business is fundamentally a prioritization problem: your attention is a finite resource, and every meeting consumes some of it. The value of an AI meeting system isn't to let you take more meetings — it's to help you stay more focused in the meetings you must attend, be more efficient afterward, and skip the ones you should skip without hesitation.
$45/month, 12 hours saved per week, follow-up misses near zero. Those are my numbers.
Several B2B members in the Solo Unicorn Club are running similar systems. Interestingly, everyone's "meeting skip criteria" are remarkably consistent — meetings with no agenda aren't worth attending 90% of the time.
Have you ever tallied up how much time you spend each week on meetings and post-meeting processing?