- Record with Fathom/Fireflies (free tier works)
- Download transcript
- Drop into Claude/GPT
- Ask: "What follow-up email would close this deal?"
Done.
But here's where it gets disgusting...
GPT remembers shit you don't even notice:
"When you mentioned John from accounting..." "Regarding your Q3 deadline you briefly touched on..." "I noticed Sarah from marketing seemed concerned about..."
Clients think you're a fucking genius.
"Wow, this person actually LISTENS"
Meanwhile you were zoning out thinking about what to eat for lunch.
The transcript caught everything.
Real example from last week:
Client mentioned their CMO "Jessica" once in passing.
I forgot her name 5 seconds later.
My follow-up email referenced Jessica's concerns about implementation timeline.
Client response: "This is exactly what we needed to hear"
The psychology is beautiful:
People drop their guard when they think you truly understand them.
Nothing says "I understand" like referencing specific details they mentioned.
Even if you were barely paying attention.
Here's my exact prompt:
"Based on this sales call transcript, write a follow-up email that:
- References specific pain points discussed
- Names the actual people mentioned
- Addresses their unstated objections
- Moves toward closing the deal"
Pro tip: Clean up the transcript first
Remove your "ums" and "uhs" Fix any transcription errors Delete the small talk
GPT works better with clean data.
The follow-up email framework it creates:
Opening: Reference specific conversation point Problem: Their exact words, not yours Solution: Positioned as "system" not "AI" Proof: Relevant to
THEIR industry CTA: Soft close based on their buying signals
Another psychological hack:
People have fragile egos about their problems.
Never say "our system is better than your team"
Say "our system enhances what your team already does well"
Same outcome. Different feeling.
What kills deals:
"Our AI will replace..."
"Your current process is broken..."
"You're doing it wrong..."
What closes deals:
"This automates the repetitive parts..."
"Enhances your existing workflow..."
"Builds on what's working..."
The transcript also reveals their REAL objections.
What they say: "We need to think about it"
What the transcript shows:
They mentioned budget constraints 3 times, integration concerns twice, and kept asking about support.
Now you know what to address.
My close rate went from 20% to 45% doing this.
Not because I became better at sales.
Because I stopped forgetting important shit.
The transcript remembers everything. GPT connects the dots.
Side benefit:
You can analyse patterns across all your calls.
"What objections come up most?"
"What features do people actually care about?"
"Where do deals die?"
Feed 10 transcripts to GPT. Get a masterclass in your own sales process.
Tools I use:
Recording: Fathom (free)
Transcription: Built into Fathom or FireFlies
Analysis: Claude for nuance, GPT for structure
CRM: Notion (with transcripts attached)
Total cost: $0-20/month
The bottom line:
I'm shit at sales. Can't remember names. Zone out constantly.
But I close deals because:
- Everything gets recorded
- GPT catches what I miss
- Follow-ups reference exact pain points
- Clients feel truly heard
Stop trying to be a better salesperson.
Start being a better listener.
Or just record everything and let GPT do the listening for you.
Your commission checks won't know the difference.
โข โข โข
Missing some Tweet in this thread? You can try to
force a refresh
99% of AI builders are competing on features while smart operators are selling $47K "chatbots" to lawyers who can't even PDF a document.
The real AI goldmine isn't in the tech - it's in the intelligence gap.
A thread on how people are printing money from digital stupidity ๐งต
Just analysed 500+ automation projects from the last 6 months.
The pattern is insane:
Companies are sitting on MILLION DOLLAR problems while asking for $2K solutions.
They literally don't know what they don't know.
What businesses think AI does:
โ Chatbots that break after 3 questions
โ Email templates with {firstname}
โ Copying data from spreadsheet A to B
โ "Hey Siri but for our company"
OpenAI just admitted AI agents will replace 92 million jobs by 2030.
But they buried the real story:
170 million NEW jobs are being created.
I spent 6 months researching what separates the winners from the losers.
Here are the 7 "recession-proof" skills nobody's teaching you: ๐งต
First, let's address the elephant in the room.
The job apocalypse is real:
- 77.4% of companies already use AI
- Entry-level positions vanishing daily
- Microsoft: 30% code is AI-written, then fired 40% of engineers
- IBM replaced entire HR departments
But that's only half the story...
Here's what the doomers won't tell you:
For every job AI kills, it creates 1.8 new ones.
The problem?
These new jobs require skills that didn't exist 2 years ago.
Skills your college didn't teach.
Skills bootcamps don't cover.
Skills that sound made up.