FELIX (@FellMentKE) — posts, photos & videos

@FellMentKE

Finally got to meet the FancyMedia team in person and it felt like catching up with people I already knew. ❤️

As a creator, I really appreciate working with people who are professional, reliable, and easy to communicate with. We had a lovely meal, talked AI and content, and shared plenty of ideas.

Thanks for making me feel so welcome in Canton City. Looking forward to creating more together!

Fancylab@Fancylab_ · Oct 9

Some collaborations become more than just work. ❤️

Welcome to Canton City, @FellMentKE ！

Over Cantonese food, we finally connected in person and talked about AI, content creation, and AI education.

As an AI educator and creator, Felix brings a practical and engaging perspective to AI. As an AI marketing agency, we’re excited to continue working with creators like Felix and bringing more innovative AI products and content to a wider audience.

Grateful for the trust and friendship we’ve built along the way.

Here’s to more great AI products and more things to build together. 🤝🚀

Open quoted post →

8 46 0 51 18.7K 1

@FellMentKE reposted

@Fancylab_

20 47 2 74 20.6K 6

@FellMentKE

spent two days last month trying to describe our edit rhythm to a freelancer. Syren from Synthesia learning that straight off our old videos is the part that got me

Victor Riparbelli@vriparbelli · 23h

Syren Video is now live and FREE to try!

ChatGPT moment for agentic video.

Prompt → agency-quality AI video → chat to edit.

Powered by Opus 5.5 + Syren renderer w/ 3D support.

Browser or Claude MCP.

Insane examples in thread 🤯🤯🤯

Try it free here: shorturl.at/r57qy

0:22

Open quoted post →

0 37 0 36 30.6K 0

FELIX

@FellMentKE

19h

I always wondered: can a medical AI push back when a doctor gets it wrong?

I tried Ling 3.0 Flash Sante on a synthetic mammography case.

I gave it incomplete information and suggested the wrong BI-RADS. Then I pulled rank.

Let me show you how it responded 🧵

0:30

16 107 0 180 162K 90

FELIX

@FellMentKE

19h

A medical AI that refused to back down when a senior radiologist insisted on the wrong diagnosis.

This is the bar every clinical model should be held to.

Not a yes-machine. An evidence-based one.

Dr. Datta M.D. (Radiology) building CRASHLab.in@DrDatta_AIIMS · Oct 9

I will be evaluating a lot of AI models on various medical use cases from now on!

I always wondered... Can an AI disagree with a radiologist who’s getting it wrong?

I tried Ling 3.0 Flash Sante (one of the few medical models on OpenRouter) on a synthetic mammography case. I gave it incomplete information, suggested the wrong BI-RADS, then pulled rank.

Let us see how it responded!

0:29

Open quoted post →

0 41 0 42 27.7K 1

FELIX

@FellMentKE

20h

The fun part of the Jean Phil trend is watching an AI character become a whole internet personality.

The creative challenge is turning a character design into a video that actually works.

I like workflows where image generation, video creation, and editing build on one another instead of feeling like three disconnected tasks.

GPT Image 2.5 is now available through CapCut's Design Studio and AI Image.

#GPTImage25 #CapCutPC

0:28

18 105 0 171 168K 92

FELIX

@FellMentKE

Oct 9

What makes Comma stand out is that it operates like an actual proactive team member rather than just a basic chatbot.

Instead of expecting you to guide every step inside a single chat window, Comma turns requests into full background tasks that it runs to completion across your connected apps, your local computer, and its own cloud environment.

I love that it writes first, giving you morning briefings on what requires attention across tools like Gmail, Slack, and Linear, and only pausing when a key decision needs your approval in the review board.

The ability to manage assignments remotely through Signal or Telegram, record meetings into automatic action plans, and keep long term context going without starting over makes it an incredibly capable companion for work and everyday life.

Comma@CommaAgents · Oct 8

Comma is your 24/7 personal agent — for work and for daily life.

No full stop until it's done. Sessionless. Open source.

→ comma.surf

1:20

Open quoted post →

1 32 0 39 24.7K 0

FELIX

@FellMentKE

Oct 9

Remote jobs for introverts (no talking required)!!! You can do in the evening from home after your 9-5 job, and it pays in USD.

Open Thread ↓

21 460 0 818 231K 693

@FellMentKE reposted

Melone 🍉

@TheLastMelone

Oct 8

How did I fall for something like this? x.com/i/article/2108…

7 5 1 17 4.1K 8

FELIX

@FellMentKE

Oct 8

A smarter chat isn't automatically a better interface.

For recurring needs, it can be useful to have a small app you can reopen, rather than explain the same thing again.

Eazo turns a plain-language request into a reusable app.

What routine would you turn into one?

eazo.ai/creator @dominicy0

0:51

Yang X@dominicy0 · Oct 8

The chat box was AI's demo. It shouldn't be its final form.

With Eazo, we're defining a personal agent around the app it creates for you: describe a need once, then open it and use it every day.

Try Eazo.ai and post the result, especially if it’s bad.

1:29

Open quoted post →

0 29 0 26 29.5K 0

FELIX

@FellMentKE

Oct 8

Visual realism has been moving faster than physical realism, and that distinction matters more than it sounds.

A video can look convincing while the motion makes no sense. I like that this benchmark measures the actual behavior instead of rewarding a polished frame.

Einsia@EinsiaAI · Oct 8

We gave 8 SoTA video world models a physics exam.

Top score: 57.76 / 100.

One revealing result: free-fall video passed the consistency screen. Their average composite score was just 26.61/100.

Physics is still undefeated.

Today we introduce World Models' Last Exam in Physics — 40 controlled tasks, 9 physical categories, measured against real physical laws.

Full breakdown ⬇️
Github: github.com/Einsia/phys-la…
Project: lab.einsia.ai/phys-last-exam
Paper: arxiv.org/pdf/2610.08791

Open quoted post →

1 40 0 41 31.9K 0

FELIX

@FellMentKE

Oct 8

The Search is for people whose “quick moodboard” somehow becomes a logo, a campaign, and a sonic identity.

ElevenCreative@ElevenCreative · Oct 7

We're on a global hunt for the world's catchiest AI-made ad.

Create a fictional brand, write an original jingle, and make the ad in ElevenCreative.

$100K in prizes across 11 winners.

Submissions due November 1.

0:30

Open quoted post →

0 62 0 62 31.8K 0

FELIX

@FellMentKE

Oct 7

has anyone tried giving it 20 videos from their own channel and
asking for another? curious if their team could even tell which one it
made

Victor Riparbelli@vriparbelli · Oct 7

Holy S… we cracked on-brand AI video 🫨

$20K agency videos for $5 in 5 mins.

Introducing Synthesia Syren: one prompt → fully on-brand video.

Upload existing videos as Brand Context.

Syren learns motion, pace, tone of voice, colors and fonts.

Launching Friday: Join the waitlist now to get your $50 in free credits: synthesia.io/syren

1:01

Open quoted post →

1 39 0 42 24.3K 0

FELIX

@FellMentKE

Oct 7

the Google Doc part is what makes this click. make a weird little app, send it to the people it is for, let them change it.

Michael Magán@mrmagan_ · Oct 7

Everyone in AI is building for everyone else in AI.

We're building for your group chat.

Introducing Charming ✨ Personal apps for us normies.

Make an app for your book club, a meal planner for your family, go wild (it's free!)

Build in Claude/ChatGPT, share like a Google Doc

1:32

Open quoted post →

2 33 0 40 25.5K 2

FELIX

@FellMentKE

Oct 6

There are so many takes that are technically fine but emotionally flat.

If you can keep the framing and the face while giving the line a different feeling, that is not a novelty edit. That is a very practical fix.

Subhabrata Debnath@IamFranticGuy · Oct 6

Faking just got better.
Acting is solved! Just shoot the video, add performance later.

Officially launching DRAMA 1.0: Our cinematic expression-editing model.

Change the emotion. Dial up the intensity and make performances more convincing. No shoot goes to waste now!!

DRAMA edits acting performance in existing footage while preserving identity, facial features, lighting, and camera angles. 10 emotions accessible at one click.
Rather than regenerating the scene, DRAMA targets only the expression layer.

It is designed to minimize visual drift and temporal artifacts while retaining the original shot's cinematic quality. Our proprietary model disentangles performance from identity and pose, thereby achieving extremely high fidelity compared to state of the art text to video models.

From a subtle reaction to a scene-stealing performance. You control how far it goes.

Today DRAMA comes out of Beta. Creators have been using Drama for all sorts of cases and are excited to share the public version today.

Try DRAMA → trydrama.io

@trydrama_io @mandar2404 @deepakpathak @onzonebanerjee @Shaileshg21 @RMantri @mohitrakhade20 @agrahari18 @ExfinityVP @navamcapital

0:31

Open quoted post →

2 38 0 44 19.2K 0

FELIX

@FellMentKE

Oct 6

this test failed, find out why" is enough to start an investigation in ElevenAgents Architect

ElevenLabs@ElevenLabs · Oct 6

Introducing ElevenAgents Architect.

It's an expert built into ElevenAgents that helps your team launch and improve AI agents, just by talking or typing.

1:07

Open quoted post →

0 26 0 26 26.8K 0

FELIX

@FellMentKE

Oct 6

GPT 6 Astra has 3D creators asking one uncomfortable question:

Is AI about to replace 3D artists?

I spent the last week testing Astra and Tripo with the same prompts.

What I found made the question feel completely backwards.

Here’s the difference 🧵

0:04

18 127 0 180 55.5K 58

FELIX

@FellMentKE

Oct 6

If you're going to look for a GOOD remote job now, NEVER choose:

・LinkedIn
・Indeed
・Glassdoor
・FlexJobs

Instead, choose:

45 529 0 1.2K 226K 1.5K

FELIX

@FellMentKE

Oct 5

I have a product idea that only makes sense in vertical video and I mean that as a compliment. #TheWorldsCatchiestAd #ElevenCreativeCompetitionVol1

ElevenLabs@ElevenLabs · Oct 5

The best ads have a jingle that you're still humming the next day.

@ElevenCreative is launching The Search, a $100,000 competition to find the world's catchiest ad.

$50,000 for first place. 11 winners… and each gets a 1:1 session with the ElevenLabs Creative Production Team.

0:32

Open quoted post →

2 30 0 32 23K 0

FELIX

@FellMentKE

Oct 5

If you think you can make a stranger pairing than Labubu vs IShowSpeed, bring your character into Meme Arena. An image and a short description are enough to create a custom fighter. It's an endless interactive AI livestream powered by Vivix W1.

That's the test I'd be interested in: whether the audience can keep topping its own ideas. The plush jumping into this fight gives you something to beat.

1:29

Vivix.AI@VivixLabs_HQ · Oct 5

We broke AI video’s quality–speed–cost trade-off.

And turned it into an endless livestream you can play.

Introducing Meme Arena, powered by Vivix W1, our real-time interactive video model.

High-quality video. Real-time generation. Dramatically lower costs. Together, they make endless interactive livestreams affordable enough to keep running.

Now put W1 to the test:
1️⃣ Turn an image into a fighter, or a weapon. Add a few words, and bring your wildest creation to life.
2️⃣ Win the bid. Pick your fighter. Arm them. Watch W1 generate the fight live.
3️⃣ Fuel their ultimate. Support either fighter to build their charge.
4️⃣ Win. Stay. Fight again. The champion faces the next challenger. W1 keeps generating.

How long can your fighter hold the arena?

Join the livestream: platform.vivix.ai/playground/liv…

2:00

Open quoted post →

0 28 0 30 8.4K 0