Zack

Helping improve healthcare through the use of AI augmentation | Helping people understand how AI works | AI Researcher | Tweets are my own.

[Zack @DocZedd]

Extraordinary results don't just happen. They're built — one ordinary action at a time, repeated, refined, and stacked until the ordinary becomes extraordinary. Start with ordinary.

[Zack @DocZedd]

@Nicolascole77 Add to the list: excessive em dashes. The current gen models love them way too much

[Zack @DocZedd]

Took me 30 years and a Costco-sized bag of apathy to get there. The silence is loud, and my calendar's got room for all the right kinds of chaos

[Justin Welsh @thejustinwelsh]

The moment you realize nobody cares what you’re doing is the moment you can finally do whatever you want.

[Zack @DocZedd]

@Dhruvam987 false, some call the Anthropic API too

[Zack @DocZedd]

The job market might seem pretty fragmented with AI etc, but there are many good startups looking for talent. If you’re looking for your next thing + have any interest in being an expert in your field - life sciences, research, design, ops, or legal, reach out to companies that seem interesting. You'll likely find many of them more receptive than you might think.

[Zack @DocZedd]

Pretty solid read. The changes being wrought by AI or not ending professional work they are reshaping the skill stack.

[Phil Chen @philhchen]

https://t.co/oMoEAYkqGl

[Phil Chen @philhchen]

https://t.co/oMoEAYkqGl

[Zack @DocZedd]

Everyone thinks the gap between them and the person who built the thing is knowledge.

It's not. It's that the builder stopped preparing.

You're not stuck. You're addicted to the threshold.

[Zack @DocZedd]

Engineers are building contextual engineering layers into their agents, but most stop at basic memory storage and call it done.

real contextual engineering isn't storage. it's a pipeline:
→ decide what to remember
→ inject it without overwhelming the model
→ merge session notes into long-term memory without stale data

the difference between a forgetful agent and a smart one lives in those three steps.

[Zack @DocZedd]

The question worth asking: are we training our replacements, or are we training a tool that lifts the floor for everyone

[Zack @DocZedd]

If you knew that every event at work was being logged, audited, and eventually used to train AI - how should we feel about the way we show up each day?

[Zack @DocZedd]

Curious to hear from clinicians or builders who shifted from "AI will replace us" to genuinely seeing it as a tool that amplifies what they do. What changed your mind?

Was it using it hands-on? A specific workflow win?

[Zack @DocZedd]

Ways AI actually wins:

1) Be dramatically faster than existing tools
2) Be far easier than current tools
3) Be meaningfully more accurate than the baseline

Pick one. Then go deep.

[Zack @DocZedd]

The Silent Break https://t.co/rVwcl9Cfkb

[Zack @DocZedd]

The silent break:

AI rewires workflows without anyone noticing.
14 handoffs later things stop working.
No error. No alert. No crash.
The system runs. The individual outputs look fine.
The chain is just broken.

Drift is silent. So are architectural gaps.

[Zack @DocZedd]

@mnedoszytko https://t.co/9j1TeMcfw7

[Zack @DocZedd]

You get comfortable being the expert.

Then you dodge anything that makes you feel clueless again.

That's when you stop growing.

Take inspiration from those like @mnedoszytko and start building.

There has never been a better time.

[Zack @DocZedd]

If you're a clinician who wants to build, stop screwing around.

Choose looking clueless over staying comfortable
Choose iterating badly over planning perfectly
Choose the ground floor over the corner office
Choose novice again over expert forever

Short term, you'll feel stupid.

Long term, you'll be the one who actually built something.

[Zack @DocZedd]

One thing about clinicians resisting AI in healthcare:

It’s often a reasonable response.

They’ve spent years mastering complex decision-making under high stakes, where mistakes have real consequences and intuition is hard-won.

Tech optimists tend to celebrate iteration and disruption.

Clinicians are trained to ask: “What happens when this fails?”

[Zack @DocZedd]

How I stay current on AI in healthcare:

- Arxiv: research papers
- a16z: implementation trends
- Case studies: real-world wins
- X for early signals and operator chatter

Literally all you need.