I make a new animation every day now, these 18 GitHub repos are my secret
this is f*cking gold for Fable 5.2 and Opus 5.5 users
Launches. Explainers. Demos. Motion. VFX. Taste.
huashu-motion → launch films that write their own music
https://t.co/lOWA4YNA6o
anything2explainer → pick a topic, get a narrated explainer
https://t.co/yUEqO4ble6
OpenScreen → screen recordings that look like polished demos
https://t.co/K9enNVzzbG
onetake → the whole piece in one unbroken camera move
https://t.co/gzUJEhLtAV
Lemo-Opuscar → 99 film looks your agent can shoot in code
https://t.co/PREjzrl6Sk
EffectCraft → a full VFX toolkit you never pay for
https://t.co/W087JtoVr3
GSAP → timelines that make every move smooth
https://t.co/8B97Y22Tkg
Three.js → 3D shots straight from a web page
https://t.co/BgbvnRcMUt
Blender MCP → Claude does the heavy 3D lifting in Blender
https://t.co/QWCqTa15R2
ComfyUI → the AI image engine behind your shots
https://t.co/RN6kGEDwOT
HyperFrames → build in HTML, deliver as video
https://t.co/dOZZVJZopQ
Remotion → videos you can edit like a web page
https://t.co/znpMp7SpNi
Remotion Skills → your agent learns video the official way
https://t.co/eNG5a2aQJc
Impeccable → a final polish pass on every frame
https://t.co/LUVwZwOvKe
Taste Skill → gives your agent a real sense of style
https://t.co/wFIFNz2wDp
Hallmark → no AI look, ever
https://t.co/xIVofZdejW
Open Design → your own AI design studio, no subscription
https://t.co/rrLI8aUxWj
FFmpeg → exports a clean MP4 every time
https://t.co/HlJpuzR3nn
Topic → script → motion → score → export.
Bookmark this, then the guide below shows how to make your first animation ↓
options existed way before black-scholes
but there was no shared way to price one, so every trade was an ad-hoc, hand-wavy negotiation
every options trader knows the bs price is wrong. doesn’t matter. it gives everyone a starting point, so instead of “i kinda think that option looks expensive” you can say “you’re pricing in more vol than i am”
and once people can argue about a number, they can trade it
that helped turn a niche contract sold through newspaper ads into the most traded contract on exchanges worldwide
ai is going to do that for everything else
every asset. every risk. every event. every future state of the world
what’s a startup worth if its biggest customer walks?
what’s your shipment worth if it lands a week late?
what’s a drug worth if the trial works but insurers won’t cover it?
ai won’t get those prices exactly right (for now). but it’ll make them easy enough to quote and easy enough to argue with, so someone can take the other side
the last market maker will be an ai
An AI training session only matters when the employee can apply what they learned to a real decision.
Can they explain what makes it acceptable, where it needs checking and what the employee should learn from a correction?
McKinsey's attached exhibit separates personal readiness for AI from organisational readiness for an agentic future. These are different measures and respondent groups; neither directly tests whether managers can assess AI-assisted work.
The article also describes a financial institution moving from broad training toward role-based learning, peer coaching and dedicated development for managers.
If they disagree about good work, resolve that disagreement before employees are evaluated against it.
Source: McKinsey & Company, “AI is changing work. Now it has to change the organization”
@RLDI_Lamy@sulefati7@ILoveBooks786@Corix_JC@FrRonconii@Nicochan33@faryus88@ozsilverfox@jameslhbartlett@ricardo_ik_ahau@timo_vi@violetainsights@DioOmega@66Iot@chessieowl@dinisguarda@BizElevatorJH@Eli_Krumova@TechieUmar@ramonvidall @TechDIYideas @World_Data_A
If loads of people are interested in your service but they're not buying, you've got an offer problem.
That is your limiting factor today. Do everything you can to solve it and cut off the distractions.
Long hours on anything else won't move you quick enough.
A negotiating AI agent can complete a transaction while overlooking what its owner actually values.
Anthropic's Project Swap tested agents in a controlled employee book exchange. Missing information about participants limited outcomes. This was a small experimental market, not evidence of enterprise procurement performance.
The commercial question I would take from it is how preferences are elicited before negotiation starts.
A price ceiling leaves many choices unresolved: delivery timing, flexibility, support and acceptable compromises. Ask the owner to rank those trade-offs and test whether the agent can explain them back.
Evaluate the brief before evaluating the bargain.
Source: Anthropic, “Project Swap: What happens when agents trade for us?”
@RLDI_Lamy@sulefati7@ILoveBooks786@Corix_JC@FrRonconii@Nicochan33@faryus88@ozsilverfox@jameslhbartlett@ricardo_ik_ahau@timo_vi@violetainsights@DioOmega@66Iot@DHTJourneys@JA_TechFuture@chessieowl@Eli_Krumova@CatDadJames@World_Data_A@TechieUmar @TechDIYideas
BOOKMARK THIS
If your Meta ads ROAS is dead, maybe your funnel is the problem. Don't always blame your creatives
Cold traffic doesn't want a buy button. It wants a story
Use the 5-page stack:
Advertorial → people who don't know they have a problem
Quiz → people who feel the problem
Listicle → people comparing fixes
Landing page → people who know your brand
Product page → people ready to buy
Collection → people ready to buy (many SKUs)
Each page targets a different kind of buyer
With this stack in place, it's hard not to make sales
Build the full stack in 30 minutes with @ecomwize
WP Engine charged $50 a month to host WordPress. GoDaddy charged $5.
That's 10 times the price, for the same open-source software. Who pays that?
It's not just "big businesses" or "small businesses".
• Small businesses that had been hacked and never wanted to go through that again.
• One of the largest banks in the world, so its branches could publish their own content.
• One of the largest software companies in the world, for its developer-relations portal.
• One of the largest fast-food chains in the world, so local agencies in small countries could publish social content and get the language and culture right.
• News sites that could get 100 times their normal traffic the day a big story broke.
A global bank, a media site, and a small business that got hacked have nothing in common using typical demographics.
But they all valued the same thing: enterprise-grade quality with the flexibility of open source, at a price between cheap hosting and an enterprise CMS. That was our target market.
To a small business, we were the best option. To an enterprise, we were the cheap alternative.
https://t.co/gW8zLV1eAb
The fastest way to blow up your AOV & CVR has always been & will always be asking yourself: “how could I justify doubling my price?” & just doing that
So to help you out:
If you are an Indian founder, pls choose a MoR platform that gives you automated FIRA/FIRC..
Exports are 0% GST, but only if you can prove the money came in from abroad.. FIRA is that proof.
@kelviqcom provides automated FIRA/FIRC for Indian founders who need it.
“Marketing is easy.”
You write a few posts, run some ads & the leads pour in.
The reality:
> Legal says no
> Sales calls the leads junk
> The CEO rewrites your headline
> The post you loved gets 4 likes
> Your budget is the first one cut
> Everyone has an opinion on the logo
> Traffic drops 30% & nobody knows why
> Sales closes the deal & rings the gong
> A board member forwards you a competitor's ad
> You're asked to "make it go viral" by Thursday
> The CFO asks what brand has ever done for revenue
> Your best campaign gets credited to "word of mouth"
> ChatGPT names three competitors before it names you
> You defend a 12-month channel in a 90-day pipeline review
> The founder's 2am LinkedIn post beats the campaign you spent a quarter building
That's why top AI labs are paying so much for these roles.
X ALGORITHM UPDATE 🚨
X just added a second way to score posts. Today it adds everything up. The new one multiplies it, and time spent sits right in the middle.
It's switched off for now. If they flip it on, a like on a post people actually read beats a like on a post they skim.
Small accounts: write to be read 👇
another massive 40% MoM growth for https://t.co/VEeW7Q340i
get the indexer that all your indexers are reselling
quickly becoming my largest saas in the portfolio
Claude runs all our meta and TikTok ads. We set the goal to 400 purchases this month and so far we have hit 102. Not bad, might not hit the target but it's doing a great job.
It scans competitor ads, sees those winning. Then designs ads using Higgsfield MCP, and launches them.
Monitors spend, distributes budget and closes poorly performing ads.
It's the best time to be alive.
in ~2 years every fintech will sell roughly the same stack. The only difference will be the door you walked in through
> @stripe: came for payments. now lends with stripe capital and just bought parafin
> @mercury: came for the bank account. now does cards, bill pay, invoicing and payroll
> @tryramp: came for the corporate card. now has bill pay and a checking account at 2% apy
the products all end up looking the same: account, card, money in, money out, credit, yield
the wedge is what stays different: the first product decides who you get, what data you see and how cheap that customer was to acquire
stripe sees every sale, so it can lend.
ramp sees every expense, so it can hold the cash. mercury sees the whole balance, so it can run payroll
so the fight moves to distribution, whoever owns the first product owns the customer
this is incredible..
just tried treg's UGC workflow..
- It scanned trending video formats on T!ktok
- Created AI UGC creator
- Wrote the script
- Made the video
- Added the voiceover
- burned in captions
And a great video overall..
All via Claude code in my terminal..
I've done this before with other tools.. but lot more hassle and time.. and lot worse video quality..
my claude session and final video 👇
Hot take: I WANT my AI assistant to have a personality.
Alex Stamos argues AI should help us without pretending to be our friend.
But why should the future of AI be some soulless corporate chatbot?
I want an AI that knows me, jokes with me, and actually feels natural to talk to.
Am I alone here?
A lot of RSI papers coming out lately.
Most of them are headlines.
Here's what we actually built.
A loop that trains a model.
Tests it on tasks it has never seen.
Keeps it only if it beats a bar we wrote beforehand.
No human decides along the way.
It's run 9 generations.
It kept 1.
That one took Google's Gemma from 37% to 60% -
On fixing broken infrastructure code.
Coding didn't get worse.
Safety didn't get worse.
And it's still not RSI.
The model got better at the task.
It didn't get better at getting better.
We wrote the loop.
We picked the tasks.
We set the bar.
The model just runs through it.
RSI is when the model starts rewriting the loop.
I haven't seen anyone show that yet.
Including us.