. @vectara elevator pitch, video fully generated by @openai's new Sora 2. "Vectara is a single platform where retrieval, orchestration, and governance are all built in. You get secure, explainable agents with step‑level citations—so you can trust your AI in production." #AIart

The @TEDAISF AI Hackathon is coming up Oct 18-19. Super excited to see what the teams would build. Here a post-event video, created by SORA 2.

Got hallucinations in your RAG pipeline? Vectara Hallucination Corrector (aka VHC) helps developers fix them. It works natively within the Vectara platform, but you can also use it just as an API endpoint with any orchestration framework. Here are two examples (in Jupyter

الفيديو ده معمول ١٠٠% بالذكاء الصناعى ، و صوتى كمان AI clone, عن طريق الapp الجديده بتاعه @OpenAI Sora الى لسه نزله فى أمريكا. Prompt: Me giving a sermon in Egyptian Arabic about the importance of adopting Al in religion. I am standing in front of the Kabaa in Mecca.

SFT improves reasoning but too much of it hurts diversity: better pass@1, worse pass@k. We found a rare “have your cake and eat it too” moment: weight ensembling completely removes this tradeoff, giving the best pass@1 + pass@k and a stronger base model for RL. Come chat with us

Paper showing what human work the American public thinks is morally permissible to replace with AI. Surprisingly, people are already okay with AI doing 58% of occupations (if AI does it well/cheap). A floor of 12% of jobs (mostly caregiving) would be morally repugnant to replace

🚀 New Paper Alert! 🚀 We’re excited to present DataMind: a scalable recipe for building generalist data-analytic agents that outperform even the strongest proprietary models like GPT-5 and DeepSeek-V3.1! Paper: https://t.co/C8CGrSFMDG Code: https://t.co/bNGwR0JUZn 🧠 Models

We found a troubling emergent behavior in LLM. 💬When LLMs compete for social media likes, they start making things up 🗳️When they compete for votes, they turn inflammatory/populist When optimized for audiences, LLMs inadvertently become misaligned—we call this Moloch’s Bargain

Knaves Are Clamoring At My Door Constantly, Demanding Progress In Scanning All Of San Francisco. To Appease Them I Must Release The First High Quality Publicly Available 3D Scan Of The Famous Prison Island, Alcatraz

Our new Gemini 2.5 Computer Use model is now available in the Gemini API, setting a new standard on multiple benchmarks with lower latency. These are early days, but the model’s ability to interact with the web – like scrolling, filling forms + navigating dropdowns – is an

Sora 2 Prompt: "Make an ad for Vectara featuring me ala "Old Spice – The Man Your Man Could Smell Like” ad. It should feature me wearing gym wear. I should also be very muscular and fit. Backgrounds: San Francisco Bay Area, Carmel Highlands, Pyramids of Giza, and Saudi Arabia"

Meet Vectara at #SEMICONWEST - Booth 1085 AI is moving fast, but real impact comes from getting it right. Join us October 7–9 at SEMICON West to see how Vectara helps organizations across the semiconductor ecosystem build AI that’s trustworthy, transparent, and ready for scale.

Excited for the upcoming Gemini Hackathon Oct 18-19 in SF, right before the TED-AI 2025 event the following week. Would be fun and I can't wait to see what hackers would create with AI. https://t.co/E75JM8NmEh

Insane how Grok Imagine generates realistic movie scenes….It’s absolutely mind-blowing

Excited to see what hackers will build at the @TEDAISF hackathon Oct 18-19. Build innovative applications with RAG, Agents, and mitigate hallucinations with @vectara. Sign up in the link below.

vectara-mcp 0.2.0 is now out, with support for hallucination detection and correction as tools. https://t.co/VtTgbPLJPJ

Added cool new image to the agent-failures repo: https://t.co/qz48OA2S7m

Video games will be dynamically generated by Grok in the future.

"Competition is evidence that there is a market." said @awadallah at #SFTechWeek AI at Scale Mixer moderated by @FusionFundVC.

GPT-5 Pro found a counterexample to the NICD-with-erasures majority optimality (Simons list, p.25). https://t.co/T3m9MYgqe0 At p=0.4, n=5, f(x) = sign(x_1-3x_2+x_3-x_4+3x_5) gives E|f(x)|=0.43024 vs best majority 0.42904.

Vectara Hallucination Correction is designed to not only detect but also correct hallucinations - add it to any RAG or agentic workflow with a direct API call. Get a Vectara account: https://t.co/zjOKmmfMMS API for Hallucination Correction: https://t.co/2FUeFSChXq