Obsessed with AI and where it's all headed
Joined April 2011
- Tweets2.2K
- Following697
- Followers28.7K
- Likes18.5K
Kathia retweeted
introducing prism, the world’s first speech-to-speech model you can’t tell from a human in conversation
prism has:
> 300ms response time, 4x faster than OpenAI GPT-Live
> outperforms every other s2s model on intelligence (with NO reasoning)
> the first s2s model customizable through reinforcement learning
we’re a team of 35 with $35M in funding, and we’re unlocking s2s model adoption faster than openai, google, and x.
the first 100 people to sign up here get 1 month free: rime.ai/prism?utm_campaign=p…
Kathia retweeted
Introducing Bivia.
A multimodal AI decision model built to complete entire workflows, not just answer individual questions.
We've spent 3+ years building and deploying AI in production across semiconductors, energy, manufacturing, and other complex industries.
Today, we're bringing that experience to everyone.
EVERY LEADING IMAGE, VIDEO, AUDIO AND 3D MODEL AT THE PRICE THE LABS THEMSELVES CHARGE. ONE API KEY.
NO SUBSCRIPTION. HOW IS THIS ALLOWED 🤯
Kathia retweeted
This is the AI feature I didn't know I was waiting for.
Not another chatbot, just something that deals with the calls I've been dodging for weeks.
ChatGPT can make phone calls now.
Setup takes 2 minutes. Add Bland as a connector in the ChatGPT plugin marketplace, sign in, done.
Every new Bland account gets free credits, so your first calls from ChatGPT are on us. No card needed.
If you want a dedicated number, unlimited calls and texts, transfers, etc. Checkout out Agent Phone Plan
bland.ai/ai-phone-number
Kathia retweeted
holy shit. Axiom just formally verified the largest proof in math history
the Enormous Theorem (CFSG) by the numbers:
1. 15,000 pages
2. 500 papers
3. 100 authors
4. 30 years of work
5. 0 humans who have read all of it
AxiomProver checked every single step in Lean in 24 days.
~40M lines of machine-checked proof, 3x the size of Fermat's Last Theorem 🤯
🤝 Paid partnership
Kathia retweeted
2024: local ai was for 4 nerds with gpus at home
2026: 7M+ downloads and a whole company launching around it
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
Kathia retweeted
POV: you open X expecting a new GTA VI reveal and it's a SaaS launch
and honestly? i'm not even mad 😭
We gave you distribution, now we’re giving you conversion! 🙇
Manage DMs and comments across your fleet of organic accounts in 15+ countries.
ManyChat-style automation, included with every TokPortal account at no extra cost:
→ Reply to comments
→ Build automated reply and moderation workflows
→ Turn viewers into customers with DM automation
Enjoy our first TokPortal Direct! 🙇
Kathia retweeted
My feed: agents, MCP, which model is best this week
My clients: "can you send the deck by Friday?"
Templafy Document Agents finally connect the two.
> built on your brand and content
> shared with the whole team
> 2 hours per deck → 8 minutes
PowerPoint just got its upgrade.
🤝 Paid partnership
AI that knows your company’s content and branding? 👀
With Templafy Team, build and share AI Agents that create editable PowerPoints using your company's knowledge and templates.
Get started 👉 templafy.com/templafy-teams-…
Kathia retweeted
a swarm of AI agents just verified the longest proof in math history
they assigned tasks, reviewed each other and merged on GitHub
zero humans in the loop. this is the future of software
Kathia retweeted
Security moving into the pull request itself is the story here.
CodeRabbit Secure opens the fix as its own PR with the attack path documented, and a person still signs off.
Kathia retweeted
NO INTERPRETER. NO ONE ASKED WHAT WAS IN HER BOTTLE. SHE DIED AT 9 MONTHS OLD.
She was born healthy in England. Her mom had arrived from Pakistan just a couple of months earlier and didn't speak English.
→ Prenatal appointments: no interpreter, against guidance
→ Two visits from a community nurse: no interpreter
→ The feeding guide she got was English-only. It never said babies under 1 shouldn't have cow's milk
→ A guide in Urdu that explained exactly that already existed. Nobody gave it to her
So her parents gave her cow's milk. They believed it would do her good.
Under age 1, cow's milk can block iron absorption and cause anemia.
Visit after visit, the notes said the same thing: "breast and bottle fed." Nobody asked what was in the bottle.
→ August 6, 2024: she's pale, her belly is swollen. She's sent to the hospital. They find a urinary tract infection and prescribe antibiotics
→ September 6: she's back. Profound anemia. They give her a blood transfusion
→ Her heart stops. They can't bring her back
She died on September 7, 2024.
The coroner called it "a lack of professional curiosity." One witness put it simply: "'bottle-fed' does not explain what was in the bottle."
After her death, the hospital made "What is in the bottle?" a standard question on its pediatric admission forms.
One question, in her mom's language, could have changed everything.
Your phone can translate a conversation in real time. Her mom went through prenatal care and two home visits without an interpreter.
Every family deserves to be asked the next question, in a language they understand.
🤝 Paid partnership
Kathia retweeted
TokPortal just put ManyChat on every account. for free.
what that unlocks:
1. auto-reply to DMs and comments
2. DM + comment automations across your whole fleet
3. TikTok and Instagram, same setup
add Claude + their MCP and you basically have an international army of accounts running 24/7
We gave you distribution, now we’re giving you conversion! 🙇
Manage DMs and comments across your fleet of organic accounts in 15+ countries.
ManyChat-style automation, included with every TokPortal account at no extra cost:
→ Reply to comments
→ Build automated reply and moderation workflows
→ Turn viewers into customers with DM automation
Enjoy our first TokPortal Direct! 🙇
Kathia retweeted
Releases like this don't come around often.
Liquid AI just dropped LFM3: a family of ~2B open-weight models covering text, vision, audio, omni and decision tasks, all built on a single shared backbone.
LFM3-1.7B was trained to work inside agent harnesses like Hermes Agent and Pi, and it handles 256K context.
In Liquid's own evaluation on WebArena-Verified, it hits 38.4, compared to 28.9 for Qwen3.5-9B. And you can fine-tune these models on one GPU.
Big potential here for local agents that run 24/7.
🤝 Paid partnership
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵
Kathia retweeted
AI just found a bug in a math proof that experts missed for decades
then it fixed it
the experts who checked the fix are in their 80s
🤝 Paid partnership
Kathia retweeted
POV: your AI agent runs 24/7 on your laptop and costs you $0 in tokens
that's basically what LFM3 from Liquid AI unlocks
before: rent a giant model in the cloud, pay per call
after: a 2B model on-device, fine-tuned on your own data, faster than everything in its class
they even ran 300K agents in parallel on the Enron emails 💀
🤝 Paid partnership
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵
I THOUGHT AI MATH WAS MOSTLY HYPE
then i read how Axiom actually got here
→ at the start of this year, one correct 10,000-line formal proof took them weeks of coaxing
→ they rebuilt the models, the harness and the infra from scratch
→ build times went from 11 hours to under 10 minutes
→ now their agents just verified a proof 4,000x that size
the longest proof in math history, checked line by line in Lean
genuinely curious what they formalize next
🤝 Paid partnership
Kathia retweeted
The largest math proof in human history is now computer-verified.
Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the "periodic table of symmetry."
About 15,000 pages, built by over 100 mathematicians across generations.
Liquid took a 2B model, fine-tuned it for a few hours on ONE GPU, and it started catching wind turbine damage that GPT-6 Luna kept mislabeling
~3,000 drone photos. that was the whole dataset
this is how i’d use LFM3:
pick one narrow task you repeat every week
collect a few thousand real examples
fine-tune it on a single GPU
run it on your own laptop, no API bill
honestly curious what people end up specializing these into
🤝 Paid partnership
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵
Kathia retweeted
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵