Generating AI Meeting Notes Without a Bot: A Manual & Hybrid Approach for Busy Team Leaders

• Budget constraint: RM80/month × 8 team members = RM640/month, or RM7,680/year—a hard ceiling for her agency.
• Meeting volume: 12–15 meetings per week meant the cost per recorded call was high.
• Hybrid need: She wanted AI-level insight without real-time bot deployment.
• Current pain: Her team was writing fragmented notes in Google Docs and losing action items in email threads.

Sarah realised that a manual-first, AI-assisted approach could deliver 80% of the value at 20% of the cost. No bot subscription. No seat licenses. Just smart workflow design.

1. Full Manual Capture

• One team member records the call (on phone or device audio).
• They take live notes in a template (agenda, decisions, owners, deadlines).
• Post-call, they clean up and share.
• Cost: RM0. Time: 15–30 minutes per 60-minute call.
• Best for: Small, focused calls (under 5 people) with a designated note-taker.

2. Recorded + AI Transcription

• Use a free or low-cost tool (e.g., Otter.ai free tier, or local provider Airis) to record and auto-transcribe.
• Spend 10 minutes reviewing the transcript and highlighting key points.
• Extract action items manually.
• Cost: RM0–RM50/month. Time: 20 minutes per 60-minute call.
• Best for: Larger or recurring meetings where searchability matters.

3. Hybrid: Manual Notes + AI Summary

• Designate one person to take live notes.
• Upload notes (or a recording transcript) to an AI tool (ChatGPT, Claude, or Gemini) to generate a formatted summary.
• Distribute the polished summary in under 5 minutes.
• Cost: RM0–RM20/month (if using a paid AI plan). Time: 10 minutes post-call.
• Best for: Formal meetings or client calls where consistency and polish matter.

Week 1–2: Setup

1. Choose a recording method: Sarah used her mobile phone's voice memo app and Zoom's local recording (free feature).
2. Create a meeting-notes template in Google Docs (attendees, date, agenda, decisions, action items, owner, deadline).
3. Assign one senior designer, Ravi, as the weekly note-taker (rotating each month).
4. Set up a shared folder: `/meetings/[month]/[client-or-project]/`.

Per Meeting (60 minutes)

• Ravi opens the template and starts note-taking live (5–8 minutes of active typing during the call).
• The call is recorded locally (Zoom free tier).

Post-Meeting (10 minutes)

• Ravi exports the recording as an MP3 and transcribes it using Otter.ai free tier (2 hours/month limit).
• He pastes the transcript into ChatGPT with a prompt: *"Extract a summary with headings: Decisions, Action Items (owner + deadline), Open Questions, Next Steps."*
• ChatGPT returns a polished, 300-word summary in under 1 minute.
• Ravi adds the AI summary to the template and posts it to the shared folder.
• Slack notification: *"Meeting notes ready: [Project] [Date]."*

Outcome (after 8 weeks)

• Time saved: ~20 hours/month on note-taking (vs. traditional manual-only method).
• Cost: RM0 (using free tiers of Otter and Zoom) or RM15/month if ChatGPT Plus is used.
• Improvement: 100% of meetings now have searchable, structured notes. Zero lost action items.
• Team morale: Designers spend less time in admin, more time creating.

Real example: Client kick-off call, 6 July 2024, with Acme Corp.

• Recording: 47 minutes.
• Manual note-taking by Ravi: 8 minutes live, 12 minutes post-call cleanup.
• AI summary generation: 2 minutes.
• Total effort: 22 minutes for a fully documented, actionable record.
• Cost: RM0.

Sarah's Monthly Cost (Hybrid Setup)

• Option A (all free tiers): RM0/month.
• Option B (with ChatGPT Plus): RM60/month (RM7.50 per team member).
• Savings vs. full bot licence: RM580–640/month.

Time Allocation Per Week

• 12 meetings × 20 minutes average = 240 minutes (4 hours).
• Ravi rotates monthly, so the load per person per year is ~3 hours/month.
• No training required beyond one 30-minute template walkthrough.

Pitfall 1: Transcription quality is poor

• *Solution:* Use a speaker who articulates clearly. If audio quality is bad, re-record with participants closer to the microphone. Otter.ai and local solutions improve with volume—test on 2–3 calls first.

Pitfall 2: Action items still get lost

• *Solution:* Create a separate Action Log spreadsheet (owner, task, deadline, status). Update it during the post-meeting summary step. Sarah used a shared Google Sheet linked in the Slack channel.

Pitfall 3: Note-taker becomes a bottleneck

• *Solution:* Rotate the role monthly. In 6 months, every team member knows the system. Backup: if the assigned note-taker is absent, the second-most-senior person steps in (pre-assigned).

Pitfall 4: Summaries lack context

• *Solution:* Feed the AI tool a 1-line prompt: *"Summarise for someone unfamiliar with [project name]. Include background, decision, and next steps."* Takes 5 extra seconds, saves 10 minutes of confusion later.

Pitfall 5: Archive gets messy

• *Solution:* Enforce a folder structure: `/year/month/[client-or-project-code]/meeting-date.md`. Sarah found this reduced search time by 60%.

For 8–15 people (small agency or department):

• Stick with the hybrid model: manual notes + free AI.
• Cost remains RM0–60/month.
• Add a second note-taker on large calls (>8 attendees) for redundancy.

For 15–40 people (mid-size company):

• Introduce a low-cost tool like Airis (RM50–100/month) for premium transcription.
• Use 2–3 rotating note-takers or one dedicated role.
• Cost: RM50–100/month + 1 FTE partial hours.
• Value gained: near-bot performance at 10% of bot cost.

For 40+ people (enterprise):

• Revisit a full bot licence (e.g., RM150–200/month for 20–50 seats).
• Calculate ROI: if your organisation records 100+ meetings/month, a bot pays for itself via time saved and search efficiency.
• Hybrid approach is still useful for sensitive calls where a human review step is preferred.