The Stripe Moment
Stripe rolled out Link for AI agents.
On paper, it’s simple:
Agents can request approval
Agents can complete purchases
Agents can transact across millions of websites
In practice, it’s a line in the sand.
Up until now, AI has been stuck in suggestion mode:
“Here’s what you should do.”
Now it can actually do it.
That means:
Procurement without back-and-forth
Subscriptions managed automatically
Workflows that don’t stop halfway
This is the shift from assistant → operator.
And once AI can move money, it’s no longer just helping your business.
It’s participating in it.
$130 Billion… And Still Not Enough
Then came earnings.
And a number that doesn’t feel real:
$130 billion spent on AI infrastructure. In one quarter.
Microsoft is running a $37B AI business
Google saw AI-driven cloud growth spike… but hit supply limits
Amazon is accelerating again
Meta continues to double down
And yet…
They still can’t build fast enough.
This is the new constraint:
Not ideas. Not demand.
Physical infrastructure.
Power. Chips. Data centers.
AI has quietly crossed from software into something much closer to an industrial system.
The Real Shift: Agents Everywhere
While everyone’s watching the spending, the product shift is happening underneath.
OpenAI is moving from Custom GPTs to Workspace Agents.
Not tools.
Not prompts.
Systems that:
Remember context across days
Execute multi-step tasks
Operate inside Slack, Salesforce, Google Drive
At the same time, Amazon is pushing agent capabilities that feel… deployable.
That’s the difference now.
We’re separating:
Things that look impressive
From things companies will actually trust
And that gap is closing fast.
Meanwhile… AI Is Quietly Doing the Hard Stuff
While all of this plays out, something bigger is happening in the background.
Mayo Clinic built AI that can detect pancreatic cancer:
Up to 3 years earlier
73% detection before diagnosis
16-month median lead time
That’s not a demo.
That’s impact.
It’s also a reminder:
All the infrastructure, all the models, all the spending…
This is what it’s for.
The Part Where Reality Shows Up
Markets are starting to ask harder questions.
Reports that OpenAI missed internal growth targets triggered pullbacks across AI-related stocks.
Because eventually, every cycle hits this point:
Can the business justify the build?
Right now:
Investment is massive
Revenue is growing
But the gap is still wide
And that gap matters.
📌 Today’s Takeaways
• AI just started moving money
Stripe enables agents to request approval and complete transactions. AI goes from suggesting work to finishing it. This unlocks real automation.
• Infrastructure is now the bottleneck
$130B spent in one quarter, and still supply-constrained. Compute, power, and data centers are limiting growth more than demand.
• Agents are replacing apps
Persistent, action-taking systems are becoming the interface. Instead of using tools, you assign outcomes.
• AI is proving real-world value
Mayo Clinic shows AI detecting cancer earlier than humans. The shift is from capability to measurable impact.
• Markets are demanding results
Missed targets at OpenAI triggered pullbacks. The question now is simple: can revenue justify the spend?
AI Tools to Try
🧠 Amazon Q
🔗 https://aws.amazon.com/q/
What it is: AI embedded across AWS for coding, data, and operations
Why it matters: Understands your infrastructure, not just your prompt
Best for: Engineering and data teams
🖥️ Claude Desktop
🔗 https://claude.ai/download
What it is: Claude running locally with app integrations
Why it matters: AI inside your workflow, not next to it
Best for: Creators, devs, operators
💬 Anthropic Claude
🔗 https://claude.ai/
What it is: Advanced reasoning AI for writing and analysis
Why it matters: Best-in-class for deep thinking tasks
Best for: Strategy, research, decision-making
💳 Stripe Link
🔗 https://stripe.com/payments/link
What it is: Payment layer for AI agents
Why it matters: Enables AI to complete transactions
Best for: Commerce, automation, marketplaces
💻 Cursor
🔗 https://cursor.com/
What it is: AI IDE evolving into agent builder
Why it matters: Moves from coding → building systems that code
Best for: Developers, AI builders
🎙️ Wispr Flow
🔗 https://wisprflow.ai/
What it is: Voice-to-text prompting across apps
Why it matters: Faster, richer input = better outputs
Best for: Busy operators
⚙️ Nebius Token Factory
🔗 https://nebius.com/services/token-factory
What it is: Run and fine-tune open-source models
Why it matters: Control your AI stack
Best for: Advanced AI teams
🧪 AI Prompts to Try
💼 Tax Profile Builder
Create a comprehensive tax profile document that includes:
- Filing status
- State
- Dependents
- All income sources (W-2, freelance, investments, etc.)
- Standard deductions and credits I typically claim
- Any special circumstances from last year's return
Format this as a reference document I can use with AI tax tools throughout the season.🎥 Video Content Repurposing
Take this [long-form video/webinar/presentation] and create:
1) A 60-second social media clip with hook and key takeaway
2) Three different platform-specific versions (LinkedIn, Instagram, TikTok)
3) A blog post outline covering the main points
4) Five quote cards for social media
Focus on the most actionable insights from the original content.📂 Project Memory System
Help me create a project continuity system with two files:
1) RULES.md containing my communication style, decision-making preferences, and project management approach
2) MEMORY.md that you'll update after each session with key decisions, next steps, and important context
Start by interviewing me about my work style and current project priorities.💳 Agent Commerce Setup
Design a spending control system for AI agents that includes:
- Approval thresholds for different purchase categories
- Required information before any transaction
- Escalation rules for unusual requests
- A logging system to track all agent-initiated purchases
Include safeguards against unauthorized spending.A Slightly Uncomfortable, Slightly Exciting Conclusion
There was a time when AI felt like a layer on top of your work.
Something you used.
Now?
It’s starting to feel like something that works for you.
And this week, for the first time…
It started handling money.
That’s not a feature.
That’s a signal.
Because once AI can:
Think
Act
And transact
You’re no longer just improving workflows.
You’re building a new kind of operating system for how work gets done.
And whether we’re ready or not…
It’s already being installed.
🧠 If you enjoyed tonight’s deep dive, forward it to someone in your network who wants to fully grasp AI in 5 minutes per day. They’ll thank you later.
Your slightly self-deprecating, definitely human narrators,
Anicia & Shane
