Jyotirmoy Sundi

Help us make AI agents secure by design. We’re looking for contributors - reach out at info@votal.ai.

Your AI ships with vulnerabilities you haven't found yet. I built a lightweight red-team repo to probe weak spots, surface failures, and make AI systems harder to break before your users do it for you. Check it out 👇

@brettcalhounn We're the identity and guardrail layer for AI agents -proving who each agent is, gating what it can touch, and logging everything, so regulated companies can ship agents without losing sleep.

Pitch me your company in one sentence.

What if your AI product could red-team itself before users do? Built a lightweight red-team repo for probing weak spots, surfacing failures, and making AI systems harder to break. Check it out 👇 github.com/votal-ai-hq/wb…

@JonasSandbrink have you tried abliteration models huggingface.co/blog/mlabonne/… huggingface.co/sundi/gpt4o-di… , would need use vllm for inference, I can help setup a smoke test if interested

I’m looking for a co-founder. Frontier labs are over-refusing legitimate bio work because they can't tell who's a real researcher and who isn't. I’m solving this with BioTrust — building the credentialing layer between frontier AI labs and researchers. Strong early interest from frontier labs. Backed by funder Sentinel Bio. I am looking for a builder who's run the trust and security backbone at a company in identity, fintech, compliance, health data, or cybersecurity. Someone with strong instincts on the trust-vs-friction calls. Finisher energy. Deeply motivated by mitigating catastrophic risk from AI and biology. Full context and how to reach me here: docs.google.com/document/d/1UB… Tag the people you know who'd be unreasonably good at this.

do & learn, do & learn, do & learn, there is no failure - very powerful, thank you @chamath

gpt-5.5 handles system design nuances really well, want to stick with it but keep hitting rate limits. switch back to opus-4.7 and the experience falls flat. stuck in a loop

@derekmeegan this is great, will try it out

Introducing the /browser-trace skill, Give your agent 100% observability into its browser: dump network requests, DOM content, screenshots, and CDP logs into a searchable filesystem. Great for reverse engineering, autoresearch loops, and monitoring the situation ™️.

debugging skills are more important then ever to fine tune from app layer to gpu infra inference layer [if using hosted models]

100% agree @nikesharora @PaloAltoNtwks That’s exactly what we built at Votal.ai. Continuous Adaptive Red Teaming (CART) autonomous AI agents that red-team your models & agents like real adversaries, 20× faster than humans (RLHF-trained adversarial attacker + 100K+ dynamic attack catalog). Enterprise AI Guardrails real-time adversarial prompt detection at 250ms latency with 99.58% accuracy. Runtime protection against jailbreaks, prompt injection, tool misuse, data leakage, and agent chaining. We don’t just detect. We harden and prove resilience before attackers strike. This is the cybersecurity industry’s most consequential moment and we built the platform for it. 👉 votal.ai #AISecurity #RedTeaming #Guardrails #AgenticAI #Cybersecurity

Claude releasing bringing markets down, claude releasing bringing markets up , would love to see a reverse impact in an industry

@elonmusk Would love to help and red team

In the next few months, we will be doing rigorous security tests of 𝕏 Chat and will open source all the code

@mikealfred thanks for sharing mike

Comments from Cantor on the recent pullback in the AI/HPC. Summary: You are simply not bullish enough.

@mikealfred Love you Mike

@derekmeegan nooooooooo

nooooo i love it when my mac opens 200 apps on restart

@DGretta_Author Yes take him to number 1, I can contri too, would love to

At this rate, I'm going to offer to take Mike Alfred out to dinner at Texas Roadhouse to thank him. I offered him before, but he said it was below his standards of restaurants. long-ish $CIFR

@mikealfred love you man, keep doing you and so happy for you, don't forget us :)

Join us again this year at the AxEC Community Tech Talk '25 - an evening of bold ideas, real conversations, and meaningful connections. 📅 Thursday, Oct 16 | 4:30–9:45 PM, Santa Clara 🔗 Signup here: luma.com/q2u0ew1w

Most think “AI red teaming” = jailbreak prompts. Reality check: 🔴 Prompt injection chaining ⚔️ Tool & API abuse 🧬 Data poisoning 🛰️ Reasoning path hijacks Votal AI simulates them all — mapped to MITRE ATLAS + OWASP Top 10. Train before the exploit. @sundi133 @Bobby_Gupta

We're thrilled to launch our AI-powered cybersecurity suite: AI Cyber Range, AI Red Teaming, AI Phishing Simulation! In 2025, with cyber threats surging 35% yearly & AI regulations tightening (EU AI Act), our solutions empower businesses to stay ahead. See more: cybersecurity.votal.ai #Cybersecurity #AI @cyber_votalai

Imagine a future where AI is your personal coach! Votal.ai's vision for AI-driven interactions is inspiring, specifically their Vive AI app, which could revolutionize personal growth. Download from Apple Store apps.apple.com/us/app/vive-da…