Happy Friday!

This week, I'm pondering two ideas. Both give a glimpse into our AI-assisted future, but in completely different ways.

"An exoskeleton, not a butler"

​Podcaster Copilot (Jake Dahn, Shruggingface.com)​

Podcaster Copilot is a proof-of-concept spawned from a tweet

Peter van Hardenberg @pvh On AI, I want an exoskeleton not a butler. 2:17 PM • Feb 1, 2024 1 Retweets 24 Likes Read 1 replies

​

The core idea here is profound:

I don't want my AI to wait for me to instruct it. I want it to see what I see, hear what I hear, and help me navigate the world.

Question: What would an 'AI Exoskeleton' look like in the context of your work?

Human-level analysis at scale

​Publicis Made 100,000 Videos Using AI To Thank Its Employees (Roberto Orosa, DesignRush)​

CEO Arthur Sadoun delivered a custom message to each of 100,000 employees at the company.

This story is not particularly sensational from my point of view. Marketing personalization assisted by AI is not a new idea.

This makes me think about what it means to have human-level analysis applied at scale.

For example: in US Healthcare, there is a function called "Clinical Documentation Integrity". To vastly simplify a complex program, doctors often get paid based on how sick their patients appear on paper, not based on what services they provide.

• "How sick patients appear on paper" = What specific diagnosis codes appear on the bill

There is strict regulation that a diagnosis cannot appear on the bill without certain criteria being met (either diagnostic results like labs or imaging, or physician actions like evaluation and management).

The role of Clinical Documentation Integrity (CDI) basically boils down to a nurse reading the patient's chart and confirming that the highest/best/most specific diagnoses were applied to this patient based on the diagnostics and physician actions documented in the chart.

CDI is highly labor-intensive AND involves activities that lend themselves well to AI...

• See diagnosis code
• Query specific guidelines for when this diagnosis code can be applied
• Read documentation from healthcare encounter
• Judge whether the diagnosis code chosen is supported by the documentation
• Judge whether a different diagnosis code (one that pays more money) would be supported by the documentation
• Repeat for each diagnosis on the bill
• Communicate findings to the doctor and ask them to sign off on the updated diagnoses

If this is happening in healthcare, it's happening in your industry too.

Question: What process would be meaningfully changed if you could apply human-level or near-human-level analysis to EVERY SINGLE (x) in your observable universe?

Hit 'Reply' and let me know. I read every response.

Adam

Whenever you're ready, here are 3 ways I can help you:

​[Individual] One-off AI Coaching ​Call: Get unstuck! Solve a problem or address an opportunity with AI assistance

​[Individual] Private Coaching: A confidential coaching program to help executives use AI assistance to get calm and organized while accomplishing more than ever

​[Team] AI Accelerator Program: An interactive workshop series for teams. Accelerate your team’s AI adoption journey and start reaping the benefits of AI more fully