AI, design, automation, knowledge management — what’s the best opportunity?

Discussion & Solutions
1

Hi everyone,

I’m looking for honest opinions and advice from people who actually make money online or closely follow current market trends.

There are so many different online income opportunities being promoted today that it’s becoming difficult to separate real opportunities from short-lived trends. I’m not looking for a “get rich quick” method. I’m looking for a direction that’s worth investing time and effort into for the next few years.

A bit about me:

  • I’m a graphic designer and high-end retoucher,
  • I work with generative AI and AI image creation,
  • I’m comfortable with new technologies and software,
  • I’m interested in automation, workflows, and productivity systems,
  • I build personal knowledge management systems in Obsidian,
  • I have experience in project work, e-commerce, and process coordination,
  • I learn new tools quickly and enjoy creating efficient systems.

Rather than chasing one-off freelance gigs, I’m interested in opportunities that can grow and scale over time. I’m open to services, digital products, education, automation solutions, AI-related businesses, SaaS, knowledge-based products, or any other promising online business models.

I’d appreciate specific suggestions: if you were in my position today, what would you focus on building from scratch? Which directions do you believe have the strongest potential over the next 3–5 years?

One request: I’m not looking for a list of random side hustles. I’m more interested in your genuine opinion on which paths are truly worth the time and have the potential to become something substantial.

Thanks for any thoughtful advice

2

my main advice is to learn comfy ui

3

Alright — here’s a money angle nobody hands you straight.

And it works whether you’re a designer, a dropshipper, a bored student, or someone who’s never made a single dollar online. So don’t scroll off thinking “not for me.” :eyes: It’s 100% for you.

:bomb: The dirty secret: the way to get paid forever isn’t being the best at anything. It’s becoming the one person a business can’t fire without bleeding.

Sounds backwards. It’s not. Let me walk you up to it. :ladder:

:mouse_trap: First — the trap they keep you stuck in

There’s an army on Fiverr cranking out product pics for five bucks.

Unlimited revisions. Done by morning. Thousands of 'em, all racing to the bottom of the barrel.

That’s the trap. Everybody starts there 'cause it looks like the door in.

It’s not the door. It’s the drain. :hole:

You’ll never out-cheap a guy charging $5 — so don’t waste a breath trying. The real money lives one floor up, and getting there is dumb-simple once you see it. :backhand_index_pointing_down:

:repeat_button: The switch: stop selling stuff, start selling “you look like money”

Pick ONE kind of shop.

Don’t sell them “10 pictures.” Sell them “your store always looks like it’s selling.”

Same work — but now it’s a monthly tab. A retainer (a deal they pay on repeat, every month) instead of a one-and-done job.

Why you should care, in one punch:

:red_square: One-off client → sticks around ~2 years
:green_square: Monthly client → sticks around almost 5

Same damn pictures. The only thing that changed is how you billed it. That’s not a tweak — that’s the whole game. :bullseye:

:robot: Where the robot still shits the bed

AI’s gotten scary-good at flat, boring product shots. Cool — let it have those.

It’s still hopeless at shiny stuff :ring: — jewelry, watches, anything reflective. Gold goes plasticky, gems turn to mush, chain links melt together. Genuinely funny to watch.

That’s your corner. The harder it is for a machine, the longer your eye keeps printing money. Start where the robot trips and you’ve got a head start nobody can patch out from under you. :high_voltage:

:sparkles: Let the machine do the grunt, you stay the magic

You don’t fight the AI. You put it to work like a cheap intern. :smirking_face:

:wrench: Robot does the boring half → backgrounds, lighting, rough drafts.
:magic_wand: You do the 10% that’s the actual reason they pay.

That 10% is your moat — the thing the all-AI crowd can’t fake and the cheap crowd can’t match.

The free toolkit, if you wanna actually run this:

  • :desktop_computer: ComfyUI — a free image-robot you run on your own PC, does the heavy lifting in batches.
  • :scissors: BiRefNet — cuts the product out clean, no janky edges.
  • :artist_palette: Flux — spits out the rough drafts you then finish by hand.

:light_bulb: Real one that bit me: I saved one client’s jewelry style (a preset — basically the AI copying a saved look) and got lazy reusing it on another. Gold came out cheap-looking, nearly lost the account. Rule burned in → one private setup per shop, never share. Slower. Worth it.

:locked: The sneaky bit nobody spells out

This is the part “they” don’t put in the guru videos.

Every month, you quietly stack up the shop’s own stuff — their style, their saved AI looks, their whole growing picture-library — and you’re the one holding it. :old_key:

Two things happen at once:

:high_voltage: Your work gets faster and cheaper for you every month (you reuse your own setup).
:chains: They can’t walk away without dumping their entire look and starting from zero.

They’re not trapped in a bad way — they’re getting a library that keeps leveling up. It just happens to make you impossible to replace. :smiling_face_with_horns:

Stolen straight from how the big stock-photo giants kept the same customers locked in for decades. Now it’s yours.

:ladder: The dead-simple ladder

:3rd_place_medal: Month 0–3 → land 3 shiny-niche shops with a one-off job. Turn at least one into a monthly deal.
(Can’t land 3? Your pitch or niche is off — fix that first, nothing else.)

:2nd_place_medal: Month 3–9 → flip everyone to monthly. Build each shop’s private library.
(Keep drop-offs under 1-in-10. Raise prices yearly — you’ll lose a couple cheapskates, who gives a damn.)

:1st_place_medal: Month 9–24 → turn the one thing you do the same way for everyone into a tiny tool you can sell. Let paying clients show you what to build — don’t build it blind.

:light_bulb: Most “I’ll build a little software tool” dreams die broke 'cause people build first and beg for customers after. You’ll already have the customers. Let them point at the tool. Backwards from everyone else — which is exactly why it works.

:warning: The trap hiding inside the plan

Your taste is your superpower and your cage.

Because only you can nail it, you can’t just dump the work on a helper — clients always want one more tweak only your eye catches. That’s how a “business” quietly becomes you, alone, at 2am, forever. :face_with_crossed_out_eyes:

The fix is baked into the ladder → build a tight system for the boring 90% so a helper runs it, and you only touch the 10% that needs your eye.

Plan that escape on day one — or you’ve just built a fancier hamster wheel. :hamster:

:wastebasket: Shiny crap to walk past

:slot_machine: AI model / girlfriend accounts — money’s loud, but it’s a casino. One ban and it’s gone, needs a whole team, can’t be sold. Skip unless you want fast cash over something you own.

:hammer_and_wrench: “AI automation agency” — everyone’s cousin is selling this. Use the automation on your own thing to print margin. Don’t make it the product.

:coffin: Selling templates / courses on your system — graveyard. Free guides everywhere. Keep your setup as your engine, not a shelf product.

:lollipop: simple-pimple — quit selling pictures. Pick one shiny corner the robot fumbles. Sell ONE shop a monthly “you always look like money” deal. Let the machine grind, you keep the taste, and quietly hold their whole library so leaving you actually hurts.

Two years of that and you don’t have a gig — you’ve got a thing you can sell. :money_bag:

You don’t need a degree, a following, or anyone’s permission to start. Just one shop and one sneaky habit. That’s it.

Go grab the first one. :bullseye: