Launch products now and grow your business with X Ads. @1x_tech’s NEO robot campaign shows how X can turn a product announcement into action. As @Kenny___Rose put it, X was the perfect launch platform.
Get ready for the Sorento India Premiere.
Join us live today at 11:59 AM IST
Witness the arrival of a new benchmark in premium SUV design, presence & experience.
Be among the first to witness the moment. https://t.co/xEH1sI78Rp
I chant your name
M
o
r
n
i
n
g
o
r
e
v
e
n
i
n
g,
i
t
i
s
t
h
e
s
a
m
e
You alone set right what has gone wrong
T
r
u
e
i
s
y
o
u
r
n
a
m
e,
y
o
u
r
n
a
m
e
Y
o
u
a
l
o
n
e
s
e
t
r
i
g
h
t
w
h
a
t
h
a
s
g
o
n
e
w
r
o
n
g
Teri sada hi Jay Ho Mata🙌
@grok
theres a 48 minute video that will make you feel like youve been using claude wrong this entire time
most people prompt it one message at a time and call it a day. this treats it completely differently, like youre onboarding a new hire instead of typing into a search bar
greg isenberg builds a real automation live on camera, a missed lead responder for a med spa, pulled straight from an idea he found on https://t.co/hqs8brsuxR. actual prompts included, not just slides and theory
the framework is nine steps. give claude a workspace like a new employee gets a desk. a brief instead of a vague ask. a ticket with an actual definition of done. eyes on your screen through desktop preview so it sees what you see. a real review process instead of blind trust. a schedule so it keeps running without you standing over it. permissions so it knows what it can just do versus what needs a yes first
ends with a seven day plan, not just theory you forget by the weekend
timestamps below if you want to skip to a specific step
0:00 intro, the ai employee map, the whole framework laid out before he builds anything
5:47 step 1, the workspace, setting claude desktop up like an actual desk for a new hire
14:53 step 2, the brief and plan mode, giving it a real brief instead of a vague one line ask
18:08 step 3, the ticket and defining done, so it knows exactly when the task is actually finished
22:15 step 4, the eyes and desktop preview, letting it see your screen instead of guessing blind
26:13 step 5, review in layers and the diff view, checking its work properly instead of trusting it outright
29:34 step 6, the schedule and routines, getting it to keep running without you babysitting every step
34:32 step 7, parallel agents and worktree isolation, running more than one of it at once without them tripping over each other
39:14 step 8, permissions, safe, ask first, human owned, what it can just do versus what needs your yes
41:15 step 9, skills connectors and hooks, wiring it into the rest of your actual stack
44:28 the seven day plan, turning all of this into something you actually do this week
47:42 closing thoughts
the gap between people getting real leverage out of claude and people still typing the same prompt every morning isnt the model. its whether they treat it like a tool or like staff
follow if you want the next one of these before everyone else finds it