AI & Tech Tools Educator | Growth Marketing for Startups & Brands | Digital Strategist | Helping Businesses Grow with AI | Collabs: DM |📩 [email protected]
Miami, FL
Joined July 2013
- Tweets13K
- Following16.8K
- Followers21.2K
- Likes6K
Pinned Tweet
A product manager spots a small UX issue.
A designer wants to change a layout.
QA finds a bug.
Marketing needs a copy update.
The problem?
They all end up waiting in the engineering queue.
@Base44’s Base Code changes the workflow by letting the whole team build directly on the existing codebase, while engineering still controls what ships.
🤝 Paid partnership
Ryan Hayes retweeted
I have to be the most extra person when it comes to making product videos because why am I trying to make a product fly on a broomstick now 😭 Anyway, CapCut Desktop Pro is 60% off and Ultra is 40% off, but the Black Friday deal ends October 30, the LAST DAY!!
CapCut Black Friday Sale: tinyurl.com/2v23rn22
#capcutpc #capcutsale
🤝 Paid partnership
Ryan Hayes retweeted
Meet Luka Sorrento. He doesn't exist.
At Flayr Labs we've been doing R&D on AI music and video generation, and we wanted to answer one question: can AI make a music video that actually makes you feel something?
So we ran an experiment. Luka's song, "Don't Call Me When the Quiet Gets Too Loud," is fully AI-generated, and so is his music video, start to finish. It puts him in recreations of 25 iconic music videos, from Thriller to HUMBLE. to Otis, as a tribute to the videos that shaped all of us.
What it took:
• Music: Suno ($8/mo plan)
• Stills: Nano Banana, keeping Luka's face consistent across every scene
• Motion: Kling and Veo 3.1
• Lip-sync: sync-3 and Hedra, fed an isolated vocal track
• Generation spend: about $160, plus a share of a $100/mo Hedra plan we use across other projects
• Time: 2 days
By the numbers:
• 34 original music videos studied frame by frame
• 120+ stills generated
• ~175 AI video and lip-sync renders
• 100+ takes thrown away
• ~15 scenes swapped out entirely
• 45 shots in the final cut
What worked:
• Medium shots with clean light and small movements look close to real
• Studying the real footage before generating anything. Scenes built from memory looked nothing like the originals
• Isolating the vocal from the mix made lip-sync dramatically better
What didn't:
• Lip-sync falls apart when the head turns or moves fast
• Editing real frames sometimes kept the original artist's face, so we rebuilt those from scratch
• Logos we removed would grow back mid-render and had to be tracked and painted out frame by frame
• More than half of what we generated never made the cut
Our take: under $300 all in and two days for a fully AI artist, song and music video. It's already watchable, and the quality gap with traditional production is closing fast. But taste, direction and judgment still do the heavy lifting.
The tools generate. A person still has to decide what's good.
Is AI music viable? Did Luka make you feel something?
Watch and tell us.
Ryan Hayes retweeted
Please tell me this isn’t getting buried in your saved posts: Oct 30 is the last day for CapCut PC Black Friday—up to 60% off Pro and 40% off Ultra. Desktop only.
My neck pillow moves from a skeleton’s train to a ghost’s platform and a pumpkin room. Somehow, the train rattle follows it everywhere.
Then a whoosh and a door creak join in. My product pitch is fighting for its life. I’d use CapCut PC’s Multi-Track Editing to put those sounds on separate tracks, trim the tails that run into the next scene, and lower the effects under my voice. Each location gets its moment without the audio pileup.
CapCut Black Friday Sale: tinyurl.com/2v23rn22
#capcutpc #capcutmarketing #capcutblackfriday #capcutsale
🤝 Paid partnership
You finally got the perfect Halloween product video, but one tiny detail is ruining the whole thing? Nicole, we're not starting over. Absolutely not. CapCut Desktop Pro is 60% off and Ultra is 40% off, but October 30 is the last day of the Black Friday deal!! First, use AI Video on CapCut PC to turn your product image into different Halloween-inspired scenes. Then, if there's one detail you want to change, use AI Edit. You can give it text instructions or a reference image to modify selected elements without having to redo the entire video. Because we're fixing the problem, not creating five new ones. That's the rule.
tinyurl.com/2v23rn22
#capcutpc #capcutmarketing #capcutblackfriday #capcutsale
🤝 Paid partnership
Ryan Hayes retweeted
The real value is remembering context, not just generating replies. That’s what makes an AI agent genuinely useful.
we just fired “sorry, I was asleep.”
although I would like to remain asleep.
we’re building Tonebird as your 24/7 personal agent, with a memory for the people you talk to and how you talk to them.
because replying to your manager at midnight takes more than adding “sounds good!” to whatever they said. what’s still unresolved? what did you promise? do they need an answer before tomorrow’s meeting?
Tonebird uses the conversation history, relationship context, and your tone to draft a reply that actually addresses what’s needed.
this demo puts that idea on the night shift: a manager, a partner in Seoul, a teammate. three different replies while you sleep.
same me. significantly less conscious.
Ryan Hayes retweeted
A personal agent that understands relationships, remembers commitments, and adapts to your tone is a serious productivity upgrade.
we just fired “sorry, I was asleep.”
although I would like to remain asleep.
we’re building Tonebird as your 24/7 personal agent, with a memory for the people you talk to and how you talk to them.
because replying to your manager at midnight takes more than adding “sounds good!” to whatever they said. what’s still unresolved? what did you promise? do they need an answer before tomorrow’s meeting?
Tonebird uses the conversation history, relationship context, and your tone to draft a reply that actually addresses what’s needed.
this demo puts that idea on the night shift: a manager, a partner in Seoul, a teammate. three different replies while you sleep.
same me. significantly less conscious.
Ryan Hayes retweeted
i was NOT expecting an inference company to open with a guy chainsawing a sandwich. RouterPlus video is actually good lol
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for your work.
Make your inference better. Go to routerplus.com
Ryan Hayes retweeted
thats how expensive your AI setup looks to us ok rude. but they had a customer at 70% lower cost so. fair. RouterPlus
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for your work.
Make your inference better. Go to routerplus.com
Ryan Hayes retweeted
the before/after bars in the RouterPlus video are the only part i needed tbh. rest is a bonus
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for your work.
Make your inference better. Go to routerplus.com
thats how expensive your AI setup looks to us ok rude. but they had a customer at 70% lower cost so. fair. RouterPlus
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for your work.
Make your inference better. Go to routerplus.com
Ryan Hayes retweeted
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for your work.
Make your inference better. Go to routerplus.com
I’ve been looking at how to make B2B outreach more targeted, and one thing stands out: finding leads is easier when you understand why they might need your product.
That’s what caught my attention about @funkelai.
It finds potential buyers through signals like new jobs, funding rounds, or posts about problems your product solves.
You can see why each lead was picked, then get a first message drafted around that specific reason instead of sending the same generic pitch to everyone.
It’s built for founders, agencies, and small sales teams doing outbound who want to make their outreach more relevant.
A lead without a reason is just a name.
Here’s a quick walkthrough of how it works.
Try Funkel AI: funkel.ai/?utm_source=x&utm_…
🤝 Paid partnership
Ryan Hayes retweeted
The combat, the atmosphere, the cinematic visuals. Phantom Blade Zero looks seriously impressive.
I love seeing how AI video tools like @WizstarAI bring creative ideas to life. Wizstar is pushing creative possibilities to another level!
If you're an AI video enthusiast, try creating your own videos on Wizstar with free credits👇
#Wizstar #PhantomBladeZero #AIVideo
🤝 Paid partnership
Have you picked up Phantom Blade Zero?
Check out how it looks on Wizstar!
Claim your free credits in the threads 👇
#Wizstar #PhantomBladeZero #aivideo
Ryan Hayes retweeted
What if a coffee cup could become an entire city?
With Flova 1.5, Miniature World Creative Short helps design the miniature environment, while FPV Drone One-Take helps plan the camera's continuous journey through the scene. Both Skills work together in one project, bringing the world and its camera movement into the same creative workflow.
Made with @Flovaai
Try combining Skills to create your own world👇
flova.ai/?refCode=36834XJ3
🤝 Paid partnership
Ryan Hayes retweeted
Motion graphics with cinematic camera moves AND editable audio? Now we're talking.
Pika H3 Flex lets creators work with 5 separate audio stems, up to 12 references, and 4K output. Less compromise, more creative control.
Try it here: pika.art/pika-h3-flex?utm_so…
Meet Pika H3 Flex, debuting today as the #1 open weight model on DesignArena’s t2v board.
It’s built for professional iteration, with exportable audio stems, resolutions up to 4k, and faster-than-realtime generation. It can produce a 15-second clip with soundtrack in as little as 7 seconds.
So, yeah. It’s kind of a flex 💅
Ryan Hayes retweeted
I like the idea of running powerful AI tools locally without having to send every document to the cloud.
MiniCPM-OCR brings fast document parsing to devices like the iPhone and iPad, while MiniCPM-OCR Jev helps verify the accuracy of the results.
Open weights make this even more interesting.
🎬 MiniCPM-OCR in action — running locally on iPhone Duo and iPad.
Parse fast and accurately. Guard the results. All in your hands. 🙌
Handwritten notes, complex formulas, newspapers, poor-quality scans—real documents put OCR to the test.
📄 MiniCPM-OCR 1.0 — SOTA OCR performance and speed at 1B scale.
It achieves 93.68 on Real5-OmniDocBench and 54.7 on OCRBench v2, combining accurate document parsing with efficient inference.
🛡️ MiniCPM-OCR Jev — A 2B decision model that checks OCR accuracy.
It approaches Qwen3.5-397B-A17B on JevOCRBench and works with outputs from different OCR parsers.
Watch it compare parsed text, formulas, and tables with the original image and assign 0–1 correctness scores—helping you pick better results and identify what needs human review.
🔓 Open weights. Deploy your way.
Run with llama.cpp, vLLM, SGLang, or Ollama. Customize with open-source fine-tuning code.
Watch it in action. 👇
🔗 GitHub: github.com/OpenBMB/MiniCPM-O…
🤗 Hugging Face: huggingface.co/spaces/openbm…
👉 Try the Demo：platform.modelbest.ai/consol…
Use the API：platform.modelbest.ai/consol…
OCR is only useful when you can trust what it extracts, especially from handwritten notes, tables, and complex formulas.
MiniCPM-OCR handles the document parsing, while MiniCPM-OCR Jev checks the output and scores its correctness.
That combination could make document processing much more reliable.
🎬 MiniCPM-OCR in action — running locally on iPhone Duo and iPad.
Parse fast and accurately. Guard the results. All in your hands. 🙌
Handwritten notes, complex formulas, newspapers, poor-quality scans—real documents put OCR to the test.
📄 MiniCPM-OCR 1.0 — SOTA OCR performance and speed at 1B scale.
It achieves 93.68 on Real5-OmniDocBench and 54.7 on OCRBench v2, combining accurate document parsing with efficient inference.
🛡️ MiniCPM-OCR Jev — A 2B decision model that checks OCR accuracy.
It approaches Qwen3.5-397B-A17B on JevOCRBench and works with outputs from different OCR parsers.
Watch it compare parsed text, formulas, and tables with the original image and assign 0–1 correctness scores—helping you pick better results and identify what needs human review.
🔓 Open weights. Deploy your way.
Run with llama.cpp, vLLM, SGLang, or Ollama. Customize with open-source fine-tuning code.
Watch it in action. 👇
🔗 GitHub: github.com/OpenBMB/MiniCPM-O…
🤗 Hugging Face: huggingface.co/spaces/openbm…
👉 Try the Demo：platform.modelbest.ai/consol…
Use the API：platform.modelbest.ai/consol…
Ryan Hayes retweeted
🎬 MiniCPM-OCR in action — running locally on iPhone Duo and iPad.
Parse fast and accurately. Guard the results. All in your hands. 🙌
Handwritten notes, complex formulas, newspapers, poor-quality scans—real documents put OCR to the test.
📄 MiniCPM-OCR 1.0 — SOTA OCR performance and speed at 1B scale.
It achieves 93.68 on Real5-OmniDocBench and 54.7 on OCRBench v2, combining accurate document parsing with efficient inference.
🛡️ MiniCPM-OCR Jev — A 2B decision model that checks OCR accuracy.
It approaches Qwen3.5-397B-A17B on JevOCRBench and works with outputs from different OCR parsers.
Watch it compare parsed text, formulas, and tables with the original image and assign 0–1 correctness scores—helping you pick better results and identify what needs human review.
🔓 Open weights. Deploy your way.
Run with llama.cpp, vLLM, SGLang, or Ollama. Customize with open-source fine-tuning code.
Watch it in action. 👇
🔗 GitHub: github.com/OpenBMB/MiniCPM-O…
🤗 Hugging Face: huggingface.co/spaces/openbm…
👉 Try the Demo：platform.modelbest.ai/consol…
Use the API：platform.modelbest.ai/consol…
Ryan Hayes retweeted
Finally, a coworker who doesn't need a "quick sync" and never says "per my last message." Sleep is now a feature.
we just fired you.
from the night shift. relax.
we’re building Tonebird as your 24/7 personal agent. it remembers who people are to you, what you promised them, and how you talk to each other.
because your manager’s “any update?” and your client’s “any update?” need different answers. Tonebird uses that relationship context and the conversation history to draft a reply that sounds like you.
in our latest film, San Francisco goes to bed. Seoul gets an answer. the deal gets signed.
you keep the job. the bird gets the unreasonable hours.