Just tried out Fellow's new AI-generated meeting summaries. I have to say, my initial reaction is... underwhelmed. It feels like it's skimming the surface for keywords and assembling generic sentences. For a tool that's supposed to be about accountability and action items, this seems like a distraction.
My main gripe: Where's the cost context? In a FinOps or engineering meeting, the critical takeaways are often tied to spend implications.
* "We're scaling up the cache cluster" → What's the projected monthly cost delta?
* "The team approved the migration to GP3 volumes" → Show me the savings estimate from the last report.
* "We decided to commit to a 1-year RI" → Which instance family, and what's the break-even point?
The AI summary I got listed these decisions as bullet points, but with zero financial linkage. It's just a transcript dressed up.
Has anyone else found a practical use for this feature in a cost-conscious environment? I'm wondering if I'm missing a configuration trick. Right now, I'd rather have a script parse my calendar for meetings tagged "FinOps" and pull the last week's cost anomaly alerts into a pre-meeting note. That would be more actionable.
```python
# Pseudo-code for what feels more valuable than an AI summary
def generate_finops_pre_read(meeting_title):
cost_anomalies = get_anomalies_last_7_days()
scheduled_ris = get_pending_reservations()
budget_burn = get_this_month_burn()
return f"""
**Pre-read for {meeting_title}**
> Anomalies to discuss: {cost_anomalies}
> Scheduled RI purchases: {scheduled_ris}
> Current monthly burn: {budget_burn}
"""
```
Is the feature just not built for our use case, or is it genuinely not parsing discussions for numerical/commitment data yet? The potential is there, but right now it feels cosmetic.