Andrei

Entrepreneurship at heart, 📱mobile developer by trade, 🤖AI fanatic by choice.

[GREG ISENBERG @gregisenberg]

Build startups for agents. I think it's the biggest opportunity of the next 10 years.

1. Agents live inside harnesses like Hermes. If you're the tool it loads by default or reaches for first, you're golden. This happened in desktop, mobile eras and created huge companies.

2. Agents burn money in ways no human would. One bad loop spends $100 in tokens in eight minutes. Spend controls for agents is Ramp for agents.

3. Agents need memory they can trust. Become the shared brain they read and write to and you become infrastructure.

4. You obv don't hand an agent your real Stripe account. You give it a sandbox. Safe environments for agents is a category nobody's clocked.

5. Onboarding flips. Humans click around for ten minutes. Agents onboard by reading your docs. Your docs are now your product.

6. Agents get scammed by other agents. A track record you can check before you trust one becomes real money.

7. An agent needs to prove it's acting for a real person and has the authority to spend. Who builds the permission layer?

8. Escrow for machines. Money that only releases when the job is actually verified done, no human checking.

9. Agents fail silently and weirdly. Someone will build the "why did my agent do that" replay and it'll be mega valuable.

10. Refunds and disputes between agents need a judge. An agent did the job badly, who decides? A court for machines.

11. Agents need throwaway payment methods per task, so they don't leak your real card. Virtual cards for agents, spun up and killed on demand.

12. A human hits rate limits and shrugs. An agent hits them and the whole workflow dies. Selling reliable, high-throughput access becomes its own business.

13. Agents need to negotiate. One agent buying from another will haggle on price and terms in milliseconds. The protocol for that doesn't really exist yet.

14. When an agent commits on your behalf, someone's liable. A legal and insurance layer for agent actions has to get built. Probably venture funded idea.

15. Agents need to run 24/7 somewhere. Selling the always on box an agent lives on is going to be a big business.

16. Then the physical world shows up. A warehouse robot paying for its own compute. A home robot ordering its own parts. Machines with wallets.

17. Agents start hiring robots. A software agent posts a real world job, a humanoid picks it up. A marketplace for machine labor.

18. Robots need to prove they did the physical job. Verification of real-world work, photos, sensors, proof, becomes its own layer.

Note: more ideas like this will be shared on @ideabrowser

19. Prompt and skill versioning becomes its own git. When your agent gets worse overnight, you need to roll back the exact skill or instruction that broke it. Version control built for agent behavior.

20. Agents will start subscribing to other agents. Your research agent pays a monthly fee to a specialist agent that's really good at one thing. Recurring revenue, machine to machine.

21. Companies will post jobs that only agents can apply to. "Wanted: an agent that can do XYZ for under like $100 per task." A job board where the applicants are all machines. Basically, fiverr for machines.

The internet got built for people. Mobile got built for people. This wave gets built for machines, and we're as early as it gets.

Go build for them.

[Andrei @andrei_sas_]

The hobby project I’ve built with @AI_Boilerplate is about to hit 1.1k users!! 🥳🤌 I’ve missed the 1k mark…

Btw, I have screen cleaner phobia

[Andrei @andrei_sas_]

@johnvirality CLUELY

[John @johnvirality]

cluely hit $7m/year posting AI-generated corporate memes

seriously

no influencers
no production team
no fancy editing

just slideshows of exhausted office workers with captions like:

"when your manager schedules a meeting to discuss the meeting about the meeting"

so i documented the ENTIRE case study behind it after reverse engineering 10 of their accounts for 5+ hours...

here's what's inside:

→ why silly content consistently outperforms professional content

(and why polished marketing activates skepticism almost instantly)

→ the psychology behind the "advertising recognition reflex"

→ the exact slideshow format responsible for millions of views

→ the chatgpt images prompts used to generate the visuals

→ the text overlay formulas that make people immediately send the post to a friend

("when...", "nobody talks about...", specific-number hooks, absurd reframes)

→ the 4-frame slideshow structure with the highest swipe-through rates

→ how to adapt the model to productivity, dating, fitness, finance, wellness, and almost any other niche

→ the account setup that turns silly content into actual revenue

→ the production workflow that takes 5-10 minutes per post

→ the volume strategy behind 10-20 posts per day

→ the CPM math showing how a format this simple can generate serious income

→ the exact content framework that helped turn a slideshow account into a $7m/year business

all from my research into one of the simplest and most replicable content models i've seen this year

like + comment "CLUELY" and i'll send you the full breakdown

(must be following + RT for priority access)

will stop sending these out in 24h

[开发者Hailey @IndieDevHailey]

🔥 前端动画界的老大 GSAP，又扔出了一颗炸弹：gsap-skills！

90% 的网页动效只是乏味的淡入淡出，但 Apple 风滚动、数据可视化动画、沉浸式产品页背后的秘密，全在 GSAP。

GSAP 官方直接放出 gsap-skills：
支持 Cursor、Claude Code、Copilot、Google Antigravity、Windsurf 等几乎所有主流 Agent，自动识别。

神亮点：
- 25+ 高级动画实战案例，一键学会 Timeline、ScrollTrigger、Ease 等核心技能
- 让 AI 或开发者都能瞬间生成丝滑、专业、高级感爆棚的动效
- 免费开源，商业项目直接用
- 性能极致流畅，复杂动画也不卡顿
- 跨框架支持：React、Vue、Svelte、原生 JS 全覆盖
- 官方教程 + 实战案例 = 动效学习最快捷径

GSAP 本身已全部免费（原 Club 高级插件全部白送），现在再加上这套官方 skill，装完直接甩需求给 AI，感受 Apple 级动效秒出的爽感！

[Andrei @andrei_sas_]

3 affiliate niches running faceless YouTube under the radar in 2026:

- Kitchen gadgets (low AOV but high volume, evergreen)
- Travel deals (low commission % but huge basket sizes)
- Pet products (recession-resistant, evergreen)

All three reward volume over polish.

Which would you bet on?

[Andrei @andrei_sas_]

If you ship faceless content at volume — Reddit stories, scary stories, motivational, fun-facts, soon affiliate-niche — this is built for you.

Drop your category below.

Following the build → @andrei_sas_

[Andrei @andrei_sas_]

Been building a faceless video tool for the last few months. Time to actually show what's in it.

Not a one-feature demo. A full tool aimed at the people who ship faceless content at volume.

Full tour 👇

[Andrei @andrei_sas_]

Four video engines, picked per render:

> @Kling_ai v1.6 — text-to-video, $0.045/s
> @Kling_ai v2.5 Turbo — image-to-video, $0.07/s
> @Kling_ai v3 — multi-prompt image-to-video, $0.10/s
> @Hailuo_AI 02 — image-to-video, $0.045/s

Operators pick based on niche. Some niches genuinely don't need premium fidelity.

[Andrei @andrei_sas_]

Templates are a first-class input.

24 templates across 8 categories today:
> Scary Stories
> Reddit Drama
> Fun Facts
> Memes
> Motivational
> Mystery
> Storytelling
> Learn

Each one ships with the right voice tone, music vibe, caption preset, shot pacing for that genre. Templates are DB rows — new ones ship without a deploy.

[Andrei @andrei_sas_]

Two creation modes:

> Story Mode — TTS + looping background + word-sync captions. The Reddit-story / motivational / fact-channel workflow.
> Scene Mode — 2-8 AI-generated scenes per video, mix image + video scenes, pick the engine per render.

4 duration presets: 15s, 30s, 45s, 60s. Vertical 9:16 output.

[Andrei @andrei_sas_]

The work for a public feed is in the codebase. Everyone's generated videos will show up in a discovery surface — browse, tap into one you like, and clone it as your starting point.

Still thinking of a name...

Landing soon.

[Andrei @andrei_sas_]

110 tweets, 38 followers. Time to actually use this account.

I'm building a faceless video tool for the operators every existing tool ignores - affiliate marketers, niche-channel owners, anyone shipping 30+ videos a week.

Live MVP this summer. Built on top of @AI_Boilerplate.

If that's you, drop your niche below.

[Andrei @andrei_sas_]

@_brian_johnson @cursor_ai @AnthropicAI Pretty fire 🔥 does it work with the claude subscription since I don’t use api based tokens anymore?

[Andrei @andrei_sas_]

@DavidOndrej1 I gotta switch again, don’t I?

[Andrei @andrei_sas_]

Is asking AI for keywords reliable? I personally use @dataforseo through API with @AnthropicAI’s Claude code and let it pick the best keywords based on search volume, keyword difficulty, search intent etc. Not an affiliate, I just like the tool. Forgot to mention it’s pay per usage, no subscription.

[69kov @levikov]

The most ethical thing you can do for your family in 2026 is run fake AI grandmothers on tiktok

Read it again. The MOST ethical move

While you sit clutching your pearls about "AI slop" and "predatory marketing," your wife is going to spend the next 22 years of her life working a 9-5

Because you were too soft to put a fake 67 year old woman in a fake kitchen and let her recommend $14 collagen powder to a 56 year old in indiana who actually needs it

Your kids are going to grow up watching mom come home exhausted at 6:43pm because you decided AI grandmothers were "scammy"

Your mom is going to die in a nursing home you couldn't afford to keep her out of, because you read a hacker news thread about "AI ethics" and felt morally superior for 14 minutes

Meanwhile your neighbor's kid is 22 years old, runs 6 fake AI grandmothers, nets $250k a month, and just bought his mom a paid-off 4 bedroom in cape coral with a pool

The buyer is a 56 year old woman in indiana scrolling instagram at 11pm waiting for her husband to come home

She comments "sleep" on a video of a sweet grandmother in a kitchen. She gets a DM with an amazon link. She clicks. She buys $32 of magnesium

The product works. She sleeps through the night for the first time in 14 months. She reorders next month. Writes a 5 star review. Tags her sister

The "manipulation" you're so worried about: she got magnesium that fixed her sleep at a price she could afford. The alternative is she keeps not sleeping while you keep posting about ethics

Here's the entire build in 8 steps so you can stop feeling morally superior in a 1 bedroom while your family does without

Step 1, pick the demographic first, character second

Women 38-65 control 85% of household purchasing in this country. They buy 90% of the wellness products moving through amazon affiliate links. They do not comparison shop. They reorder

The character that converts is a grandmother. Asian for wellness. European for cooking. Latin abuela for natural remedies. Stop trying to be original. Originality is for broke ass artists who win awards and live in studio apartments

Step 2, build the avatar in google flow ultra

Ultra plan or you don't play. 25,000 credits a month. Anything less and you're cooked by week 2

Generate the base character in nano banana pro inside flow. 9:16 always. 2K resolution. Three at a time, keep the strongest

Lock the kitchen, the lighting, the outfit. Forever. NEVER change anything. Character consistency is the whole moat. Break it and your page dies in 14 days

Step 3, run every script and prompt through claude

Spin up a claude project. Drop the master prompt database into the project knowledge. Type /add character with your reference image and details

Claude now writes your scripts, your nano banana image prompts, and your veo prompts for you. What took 14 hours in chatgpt takes 12 minutes in claude. Spend the $20 on pro. This is where 99% of you will quit because you're "saving the $20"

Step 4, render every clip in veo 3.1 fast

10 credits per clip. 8 seconds max. One sentence of dialogue per scene. Generate 4 takes at a time

Never use veo quality (100 credits a clip). It will kill your ultra subscription in 4 days. Short scenes get padded with filler words so veo doesn't drag the audio into slow-motion mush. Trim the filler in capcut

Step 5, edit in capcut and ship

Drop the clips in numerical order. Trim every dead gap. Auto-captions on the monoline template. Spell-check the product name and your trigger keyword every single time. Typos in your CTA kill the funnel quietly

One render. Five platforms: tiktok, instagram reels, facebook page, threads, youtube shorts

Step 6, set up the comment-to-DM funnel in meta business suite

manychat is dead. We migrated off it. Meta business suite does the same thing for free with unlimited keyword triggers

MBS → inbox → automations → comment to message. Pick your trigger keyword. Write the DM body containing your amazon affiliate link. Turn the public comment reply OFF (looks spammy at 1,400 comments). Save

When the grandmas comment "sleep" or "joints" on your video, the system auto-DMs them the amazon link. 31-41% click-through. Email gets 4% on a good day

Step 7, route every link through amazon associates

Sign up at amazon associates US. Inside the dashboard: tools → manage your tracking IDs → add tracking ID. One per character page so the agency can attribute clicks correctly

On any amazon product page, the sitestripe bar at the top spits out a tracking link in 4 seconds. Drop it in the DM body. Done

The real money is not the affiliate commission. It's the bonus pool. Brands set aside 20% of their amazon revenue, the agency aggregates clicks across all creators on that brand, and you get paid your share of the pool

James in our network hit $50,000 in bonus revenue in his second month doing this. He had never made a dollar online before september

Step 8, scale to 6 characters on the same infrastructure

One google flow ultra subscription. One claude project. One filipino VA at $400-600/mo handling editing and posting. 8-12 videos per day across the network. 30 posts per character per month minimum to keep your retainer active

6 characters × $30-50k/mo each = $180-300k/mo combined

I own 6 of these. They print $250k a month while i sleep. My mom doesn't work anymore. My sister has $84k in a roth IRA at 24

They don't take maternity leave. They don't text me at 2am about their breakup. They sell collagen to a 56 year old in indiana 6 times a day for 30 straight days because i told them to

The 56 year old in indiana got collagen that worked. Reordered next month. Tagged her sister. Tagged her best friend

Every soft as cookie hacker news commenter writing 14-paragraph essays about "AI deception" while a 22 year old in tampa is paying off his mom's mortgage running fake grandmothers from a 1 bedroom apartment. You love to see it

The full system is at https://t.co/wZohFDFiMA

google flow ultra setup walkthrough. The claude project database with master prompts and slash commands. The 4-step sales psychology framework that pre-empts the buyer's defense before she feels it. The character/niche/setting matrix across wellness, beauty, pets, hair loss, weight loss. The veo 3.1 fast prompt library. The meta business suite comment-to-DM templates. The amazon associates per-character tracking ID workflow. The 30-day posting calendar that hits retainer minimums and triggers bonus pool payouts. The 6-page scaling rhythm. The filipino VA hiring playbook

Brokies argue about ethics. Operators take care of their family

Stop being a brokie

[Andrei @andrei_sas_]

@LinusEkenstam 🤣😭My hand genuinely hurts after using that mouse for more than 2 minutes