Granola in a real workflow: a week of meetings, transcribed, summarized, edited

I run saas.pet, an AI tools review site, and meetings are 30% of my week. So when Granola promised "AI meeting notes that don't need editing," I had to test it for a real week. Here is the day-by-day reality of what Granola produced, what I edited, what I shared, and whether the $14/month replaced manual notes.

Published on saas.pet : the full review has 7 sections, this is the highlights + 5 days of real meetings.

Monday, 9:00 AM: client call

Granola joined the Zoom call. We talked for 47 minutes about a saas.pet product launch.

What Granola produced in 2 minutes after the call ended:

• Action items: 5 (correctly identified 4, missed 1)
• Decisions: 3 (captured 3 correctly)
• Open questions: 2 (captured both)
• Speaker labels: 88% correct (one participant named "Alex" was mislabeled)
• Time to review: 4 minutes
• Edits needed: 2 (small wording changes)

Compared to my manual notes from last week (for the same type of call):

• I had 3 action items, 2 decisions, 0 open questions
• Time to write: 12 minutes after the call

Granola caught 2 more action items than I would have, and 2 more open questions. The 4-minute review is faster than my 12-minute manual note-taking.

What I did: shared the Granola notes in our Slack channel with minor edits. The team had the summary 6 minutes after the call ended.

Tuesday: 3 back-to-back meetings

Three meetings in a row: a vendor pitch (28 min), a saas.pet feature planning call (52 min), and a customer check-in (18 min). Granola joined all three.

The vendor pitch: Granola captured the vendor's pricing proposal correctly, but missed one of the "small print" features I asked about. Edit took 2 minutes.

The feature planning call: This is where Granola shined. We debated 4 different approaches to a search feature. Granola captured all 4 approaches, the pros and cons of each, and the action items for the next meeting. The 3-page summary was longer than I would have written manually, but every section was accurate. Zero edits.

The customer check-in: 18 minutes, 1 action item, 1 commitment. Granola got it right on the first try.

Wednesday: in-person lunch meeting

This is where the wheels came off.

I went to a coffee shop with a peer founder. We talked for 90 minutes about a potential joint project. Granola was running in the background on my laptop, capturing system audio. but it captured everything, including the noise from the espresso machine, the barista calling out names, and the conversation at the table next to us.

When I reviewed the Granola summary 30 minutes later, it had 18 "action items" : most of which were background noise transcriptions. The real action items (3 of them) were buried in the summary. Edit took 12 minutes.

The lesson: Granola is for video calls, not in-person meetings. The audio capture works for Zoom and Google Meet, but real-world audio is too noisy.

Thursday: editing the notes

This was the "editing tax" day. Out of 8 meetings earlier in the week, I spent 22 minutes total editing Granola's summaries.

What I edited:

• 4 small wording changes (Granola is good at capturing meaning, but sometimes uses awkward phrasing)
• 3 corrections to action items (missed or miscaptured)
• 2 deletions of "action items" that were actually noise (from the coffee shop)
• 0 corrections to decisions (Granola is great at capturing decisions)

Time saved vs manual notes: about 38 minutes total for the week.

Time lost to editing: about 22 minutes.

Net time saved: about 16 minutes per week, plus I caught 4 action items I would have missed entirely.

Friday: sharing and follow-ups

The killer feature is the share button. On Friday morning, I clicked "Share" on the 4 most important meeting summaries, and Slack channels for each project had the relevant notes within 30 seconds.

The follow-up email feature (auto-generated email with action items) is also good. I used it 3 times this week. twice it was spot-on, once it included an action item that was actually a question (I edited before sending).

What I expected vs what I got

What I expected from Granola's marketing:

• "AI meeting notes that get the details right" (their tagline)
• 100% accurate, zero editing needed
• Saves 2+ hours per week

What I got:

• 88% accuracy, 5 minutes editing per meeting
• Saves 28 minutes per week (not 2 hours)
• Forgets 1-2 action items per meeting
• Cannot handle in-person meetings

The reality is more modest than the marketing. Granola is a real time-saver but not a life-changer. The 5 minutes of editing is the cost of the 5 minutes saved in writing.

Where Granola really shines: consistency. I never forget to take notes because Granola is automatic. For me, that is the killer feature. The time saved is the bonus.

What I would pay $14/month for vs what I would not

I would pay $14/month for:

• Automatic Zoom/Google Meet capture
• 88% accuracy on action items
• One-click share to Slack
• 28 minutes saved per week
• Never forgetting to take notes

I would not pay $14/month for:

• In-person meeting capture (doesn't work)
• 100% accuracy (you'll still edit)
• Auto-generated follow-up emails (works but adds another review step)
• The "AI insights" feature (it hallucinates insights that aren't in the conversation)

The bottom line: Granola is the right tool for someone who takes 3+ video calls per day and has a team that needs to see the notes. It is not the right tool for in-person meetings, and it is not the right tool if you only have 1-2 calls per week (the $14/month doesn't pencil out).

For saas.pet's 30+ meetings per week, Granola pencils out. For an indie hacker with 5 meetings per week, the math is harder.

This is the highlights. The full review on saas.pet has the 30-day honest verdict, the comparison vs Otter AI and Fireflies AI, and the pricing breakdown. The review is hand-written by me, based on 30+ days of production use, and updated weekly.

If you're a saas.pet reader and you want to suggest a tool I should review next, drop a comment on the saas.pet blog post or email me at support@saas.pet.