Home › Blog

AI will not replace you.

AI will not replace you.

AI will not replace you.

but someone who builds with it at 3am while you were writing a 47-slide deck

about "AI Transformation Strategy"

already did.

not because they're smarter.

because they stopped asking permission.


"excited to announce we are leveraging AI"

you leveraged a text box.

I leveraged a text box.

my aging step-father could leverage a text box.

he makes better prompts than your consultancy.

"we are building an AI-first organisation"

no you're not.

you bought 400 Copilot licences

and your CISO found out on LinkedIn.

"AI will create millions of new jobs"

it will.

writing prompts for people

who can't write prompts

for people who can't write prompts.

it's prompts all the way down.


here is what nobody posts:

I mass mass mass hallucinated last Tuesday.

the model was wrong and I shipped it.

the vector db ate 6GB of ram and died at 2am.

the "semantic search" was just TF-IDF in a trench coat.

the graph had 160K nodes and the LLM summarised

three of them. confidently. with citations.

to papers that do not exist.

nobody posts that.

because failure doesn't get

the little blue thumbs up

from the VP of Digital

who just discovered ChatGPT

in February.


"what LLM do you use?"

brother I use seven.

and a graph database.

and a message bus.

and three WebSocket streams.

and a Bayesian causal engine

that still thinks Tuesdays are suspicious.

the LLM is the easy part.

the LLM is the TEXT BOX.

the system is the weapon.

one model does not a platform make.

one prompt does not a strategy make.

one LinkedIn post does not a thought leader make.


hot take incoming:

your "AI strategy" is a prompt.

my AI strategy is 8,600 lines of Python,

32 floating windows,

15 live threat feeds,

a 3D galaxy you can fly through

with WASD like it's 1999,

and a sarcastic AI assistant

who calls your CVEs mid.

built by one dev.

one graph.

one terminal.

and mass amounts of coffee.

that's not a strategy.

that's a Tuesday.


"but scott what about AGI"

mate I can't get Neo4j

to stop throwing type errors

on a Wednesday afternoon.

we are not close to AGI.

we are close to

very confident autocomplete

with excellent vibes.

and I say that

as someone who BUILDS with it.

every. single. day.

the gap between the demo

and the deployment

is where careers go to die.

and LinkedIn posts go to thrive.


so here is the bit

where I'm supposed to say

"agree? disagree? let me know in the comments"

no.

go build something.

go break something.

go ship something at 3am

that nobody asked for

that solves a problem

nobody knew they had.

then come back

and tell me about your AI strategy.


one dev. one graph. one AI.

160,000 threat entities.

zero slide decks.

ninja.ing

#CyberSecurity #ThreatIntelligence #AI #BuildDontSlide #NinjaSignal

The Probably Fine Daily

Threat intelligence every morning — new victims, new groups, what matters, in plain English. Free, with receipts.

Subscribe to the Daily →

Originally published on LinkedIn ↗

← All writing