Jed Casey

Building at

[Boris Cherny @bcherny]

Opus 4.7 feels more intelligent, agentic, and precise than 4.6. It took a few days for me to learn how to work with it effectively, to fully take advantage of its new capabilities.

Will post a few more tips throughout the day, starting with this blog post: https://t.co/XQrH8P28yo

[Jed Casey @jed_casey]

The full piece breaks down why the spending-vs-ROI gap exists and a practical framework for closing it.

Worth a read if your company is spending seven figures on AI and nobody can explain what for.

https://t.co/dkSYjREjoB

🧵 7/7

[Jed Casey @jed_casey]

https://t.co/oqtZGW4nsK

[Jed Casey @jed_casey]

The companies getting real ROI all did the same thing first. They didn't ask "what can AI do?"

They asked: "What costs us the most time and judgment right now?"

Different starting question. Completely different outcome.

🧵 4/7

[Jed Casey @jed_casey]

A simple audit every leader should run this quarter:

List your 10 most time-consuming workflows. Score each on (1) how rule-based, (2) how many systems it touches, (3) how often it happens.

Your top 3 are where to start.

🧵 6/7

[Jed Casey @jed_casey]

97% of enterprises report *some* benefit from AI. Only 29% see significant ROI.

That 68-point gap is where most companies live - using AI enough to feel productive, not enough to transform anything.

🧵 3/7

[Jed Casey @jed_casey]

Anthropic just hit $30B in annual revenue. 1,000+ companies spending $1M+ each on Claude alone.

The experimentation phase is over. This is budget-line-item money.

So why are 79% of organisations still struggling with AI adoption?

🧵 1/7

[Jed Casey @jed_casey]

Because most companies adopted AI the way they adopt software: roll it out, train people on features, hope for results.

That works for Slack. It doesn't work for something that fundamentally changes how work gets done.

🧵 2/7

[Jed Casey @jed_casey]

https://t.co/oqtZGW4nsK

[Jed Casey @jed_casey]

"Steelman the opposite position" is the single most useful prompt for strategy work. If you can't articulate the strongest case against your recommendation, you haven't finished thinking yet.

[Jed Casey @jed_casey]

Question 5: "Simplify this until a graduate could challenge it"
Complexity is where bad strategy hides. Full article with all five in depth 👇
https://t.co/lqXVHUAwWG
🧵 7/7

[Jed Casey @jed_casey]

https://t.co/i0LOmExLb2

[Jed Casey @jed_casey]

Most strategy dies in the room where nobody pushed back.
Not because the thinking was weak. Because nobody asked the right questions.
🧵 1/7

[Jed Casey @jed_casey]

Question 4: "Steelman the opposite position"
If you can't articulate the strongest case for the path you didn't take, you don't fully understand your own.
🧵 6/7

[Jed Casey @jed_casey]

https://t.co/i0LOmExLb2

[Jed Casey @jed_casey]

The people getting extraordinary results from Claude aren't prompt engineers. They're people who spend 30 minutes thinking WITH it before asking it to build anything.

[Jed Casey @jed_casey]

The people getting extraordinary results from Claude aren't prompt engineers.
They're people who spend 30 minutes thinking WITH it before asking it to build anything.
Full method in today's article 👇
https://t.co/jeolyKkl5A
🧵 7/7

[Jed Casey @jed_casey]

https://t.co/B0cEexxzr6

[Jed Casey @jed_casey]

The people getting the best results from Claude aren't writing better prompts.
They're giving better briefs.
There's a difference, and it changes everything.
🧵 1/7

[Jed Casey @jed_casey]

Step 3: Map each principle to your context. Some transfer cleanly. Some break - and those breaks reveal things about your own problem you hadn't considered.
Step 4: NOW build the deliverable. Claude has 30 minutes of shared context. Night and day difference.
🧵 6/7