Currently accepting 2 clients for Q3 2026
See if we are a good fit
Growth

this is why AI still feels useless to some people

I have been playing with BOSSMADRE, my openclaw agent, and RETARDLA, my hermes agent, a lot lately. Basically two AI operators living inside Discord.

I’ve been playing with BOSSMODE BOSSMADRE, my openclaw agent, and RETARDLA, my hermes agent, a lot lately.

basically two ai operators living inside discord.

sounds nerdy because it is.

but they help me with research, project qa, landing pages, newsletter stuff, ad workflows, reports… all the annoying things that usually sit in my head until I finally have time.

but I don’t want to make this email about “look what my ai can do”

that’s not useful for you.

the useful part is this:

AI works better when you give it a repeatable marketing workflow, not a vague question.

here’s your stupid-simple high roi insight:

pick one annoying ad decision you make every week and turn that into an ai workflow.

not your whole business.

not your whole marketing system.

one decision.

if you run ads, you already know the matrix:

open ads manager.

look at spend.

look at CPA

look at ROAs

look at CTR

look at the same 19 tabs you looked at yesterday.

then you try to decide:

“do I kill this or am I being impatient?”

lol

that decision alone is a good AI workflow.

you could automate and make sure that AI has your last 7 days of ad data and ask:

Split these ads into kill / watch / scale. Tell me why. Use spend, CTR, CPC, CPA, ROAs and creative angle. If the data is too early, say that instead of pretending.

that last line matters btw.

“if the data is too early, say that.”

because ai loves giving answers even when it shouldn’t.

same as a junior marketer who wants to look smart in a meeting.

so you have to give it rules.

this is where most people mess up.

they ask:

“how can I improve my ads?”

and then ai gives them the most basic answer ever:

test more hooks, improve your creative, understand your audience bla bla bla.

thanks bro. very helpful.

but if you ask:

here are my 5 best ads and 5 worst ads. find the pattern difference between them. then give me 10 new angles that borrow from the winners without copying them.

now we’re talking.

that’s something you can use.

same with landing pages.

bad prompt:

“improve this landing page.”

better prompt:

compare this ad copy with this landing page. where does the promise change? where would a buyer feel tricked, confused or bored?

same with competitors.

bad prompt:

“give me competitor insights.”

better prompt:

here are 20 competitor ads. group them by angle, offer, visual pattern and proof type. then tell me what they all ignore.

this is why ai feels useless to some people.

they’re using it like google with confidence.

but the better use case is ops.

tiny marketing ops.

daily stuff.

weekly stuff.

stuff that is too annoying to do perfectly, so you either rush it or avoid it.

that’s where I’m using bossmadre and retardla more now.

I’ll give them a clean task, files, context, rules, and tell them what the output should look like.

sometimes they still mess up.

ofc they do.

but when the workflow is clear, it gets scary useful.

and when the workflow is messy, it just makes a faster mess.

so if you want to steal this, steal this:

daily ad triage

  • input: yesterday’s ad data
  • output: kill / watch / scale
  • why it helps: less emotional staring at ads manager

ask:

look at these ads and split them into kill / watch / scale. use spend, ctr, cpc, cpa, roas and any creative pattern you notice. don’t be generic. explain why.

creative pattern mining

  • input: your last 10 winners and losers
  • output: what hooks, visuals, offers and proof angles show up
  • why it helps: you stop starting from blank page everytime

ask:

look at these winning and losing ads. what patterns show up in the winners that are missing from the losers? then give me 10 new angles based on the winners.

landing page mismatch check

  • input: ad copy + landing page
  • output: where the promise breaks
  • why it helps: sometimes “bad traffic” is just bad message match

ask:

compare this ad with this landing page. where does the promise change? where would the buyer feel confused, tricked, or bored?

weekly competitor swipe

  • input: ads you saved
  • output: 5 angles worth stealing and 5 to ignore
  • why it helps: your creative brain stays fed without scrolling for 2 hours

ask:

group these competitor ads by angle, offer, visual pattern and proof type. tell me what is overused, what is interesting, and what nobody seems to be saying.

that’s it.

nothing crazy.

no “ai revolution”

just taking the boring decision you already make and making it easier to repeat.

because the real question is not:

“how do I use ai?”

it’s:

“what part of my marketing workflow is slow, annoying, repeated, and clear enough to delegate?”

start there.

if you want me to look at your current ads/funnel/ops and tell you the first workflow I’d automate or go into details on exactly how I have this setup, reply to this email.

till next week, Kris

Next Letter, Next Wednesday

Want the next one in your inbox?

One letter, every week. Stupid simple, high-ROI insights for founders who need profit, not theories.