Continuing to explore Healthcare Ai and its implications from different angles, I just published my latest essay.

By the time I’m old enough for the senior menu, my android co-author will almost certainly have followed me home. Right now, that “android” is just a large language model in the cloud that helps me write. Fast-forward a couple of decades and systems like it will likely be:

* listening for changes in my voice that hint at cognitive decline

* watching my gait through wearables and smart floors

* negotiating with my cardiologist’s models about my meds* and hiding inside practical hardware: walkers, beds, lift robots, and smart homes

In my latest Substack essay, “𝗕𝗼𝗱𝗶𝗲𝘀, 𝗕𝗼𝗹𝘁𝘀, 𝗮𝗻𝗱 𝗕𝗶𝘁𝘀: 𝗔𝗴𝗶𝗻𝗴 𝗶𝗻 𝘁𝗵𝗲 𝗘𝗿𝗮 𝗼𝗳 𝗠𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗔𝗜,” I argue that aging, at its core, fails along three coupled axes:

𝗠𝗼𝗯𝗶𝗹𝗶𝘁𝘆 – can you move safely and independently

𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻 – can you remember, plan, decide

𝗖𝗵𝗿𝗼𝗻𝗶𝗰 𝗹𝗼𝗮𝗱 – how many conditions you’re carrying, and how volatile they are?And that the real opportunity for healthcare isn’t a single humanoid robot, but 𝗹𝗼𝗼𝗽𝘀 where:

* 𝗕𝗼𝗱𝗶𝗲𝘀 throw off signals,

* 𝗕𝗼𝗹𝘁𝘀 (devices, robots, environments) act and assist,

* and 𝗕𝗶𝘁𝘀 (multimodal models + LLMs) make sense of it all and talk to us in plain language.

The question I’m wrestling with is less “Can we build this?” and more:

> Will these android helpers keep us merely alive, or help us live the way we 𝘸𝘢𝘯𝘵 for as long as possible?

> And for whom will they work: a thin slice of older adults, or everyone?

If that resonates, you can read the full essay here:👉 [https://lnkd.in/gH58afNq)

Most of my worst AI product ideas started in very pretty rooms.Innovation hubs. AI labs. Cool POC demos.I assumed that if we built impressive pilots, the rest of the organization would just get it and turn them into products.They didn’t.Thus my latest 𝗘𝗹𝗲𝗰𝘁𝗿𝗶𝗰 𝗦𝗵𝗲𝗲𝗽 post: 𝟴𝟴% 𝗼𝗳 𝗔𝗜 𝗣𝗶𝗹𝗼𝘁𝘀 𝗗𝗶𝗲. 𝗛𝗲𝗿𝗲’𝘀 𝗛𝗼𝘄 𝘁𝗵𝗲 𝗢𝘁𝗵𝗲𝗿 𝟭𝟮% 𝗦𝘂𝗿𝘃𝗶𝘃𝗲 𝗶𝗻 𝗛𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲.It’s a field report from decades of building AI products, most recently in regulated healthcare, including:• Why pilots are easy but 𝗿𝗲𝗴𝘂𝗹𝗮𝘁𝗲𝗱, 𝗿𝗲𝘃𝗲𝗻𝘂𝗲-𝗱𝗿𝗶𝘃𝗶𝗻𝗴 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝗮𝗿𝗲 𝗵𝗮𝗿𝗱• The difference between an 𝗔𝗜 𝗹𝗮𝗯 and an 𝗔𝗜 𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺• How 𝗠𝗟 𝗶𝘀 𝘁𝗵𝗲 𝗲𝗻𝗴𝗶𝗻𝗲 𝗮𝗻𝗱 𝗚𝗲𝗻𝗔𝗜 𝗶𝘀 𝘁𝗵𝗲 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲—and why you need both• How internal copilots (for engineers, ops, etc.) help, but 𝗿𝗲𝗮𝗹 𝘀𝘂𝗰𝗰𝗲𝘀𝘀 𝘀𝗵𝗼𝘄𝘀 𝘂𝗽 𝗶𝗻 𝗮𝗱𝗵𝗲𝗿𝗲𝗻𝗰𝗲, 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗮𝗻𝗱 𝗿𝗲𝘃𝗲𝗻𝘂𝗲, not just “productivity gains”• What it actually takes to align a whole company—from Board / ELT to manufacturing line—around AI, including education and upskilling at every levelIt’s grounded in healthcare (sleep, respiratory care, FDA-cleared AI products), but the playbook applies to any large organization that’s tired of AI theatre and wants systems that actually ship.If you’re wrestling with how to get from pilots to products—especially in high-stakes, regulated domains—I’d love your take.

“Blood, Breath, and Bits: How AI is learning the language of the body”This one is about the convergence I’m seeing up close:·       We’re now surrounded by continuous data about our bodies and lives – sleep, heart rhythms, breathing, glucose, movement, even how we speak and type.·       New foundation models based on sensor data (think “LLMs, but trained on EEG/ECG/respiration instead of text”) are starting to learn deep patterns in that stream.·       On top of that, we’re seeing the early prototypes of personal health agents that can analyze your data, reason with medical knowledge, and talk to you in normal language.I use sleep as the anchor, because that’s where some of the most interesting work is.But the point of the piece is broader:·       The same architecture applies to most chronic conditions – cardiometabolic disease, mental health, aging, and the messy combinations many of us actually live with.·       If we connect the dots well, we move from one-off snapshots in the clinic to a richer, continuous view of the whole person.If you’re curious about this broad theme, I’d love for you to read it and, if it resonates, please subscribe.

Electric Sheep is a longform podcast about artificial intelligence — and what it is quietly doing to us.

Hosted by Badri Raghavan, a senior AI executive and physicist who has spent decades building real-world AI systems, Electric Sheep explores how modern algorithms are reshaping healthcare, work, identity, and care itself.

This isn’t hype.
And it isn’t panic.

Each episode sits inside the real tensions of applied AI: models that diagnose disease, digital agents that “listen” to patients at 2 a.m., systems that can sound more empathetic than doctors — and the social, clinical, and ethical consequences that follow.

Expect deep dives into:
• Sleep, chronic disease, and mental health
• How AI actually ships inside regulated companies
• Empathy at scale
• Governance, bias, and real-world failure modes
• What happens when simulations start replacing care

If you’ve ever felt you’ve been given a front-row seat to a transformation you don’t fully understand yet — this podcast is for you.