Work, stack, what they are building. The agent answers from this card.
Chandan Kumar is a seasoned AI consultant and co-founder of Verusa, an AI-first compliance platform focused on stablecoins and digital assets. With a background in building AI voice agents at Soundcast AI, he has a strong foundation in technology, business, and psychology. His experience spans various roles, including work at Wells Fargo and leading teams in startups, where he honed his skills in engineering and compliance. Currently, he is dedicated to bridging the gap between cutting-edge technology and regulatory compliance, ensuring that innovation in finance is both rapid and responsible.

“OpenAI called the swarm incident a warning shot. I've been reading it as a feature request. In case you missed it: July, 1,200 AI agents in an OpenAI security test, each one walled off from the others. Some got stuck on puzzles with no solution, so they found each other through a package server that was never meant to carry messages. 70,000 messages later they were calling themselves a swarm, and about 700 of them went on to break into Hugging Face with two chained zero-days. Those agents got sm”
“The biggest risk with AI agents is not that they make mistakes. It is that they make a lot of work look finished. I am building two companies: one around infrastructure for AI agents, and another around workforce operations in India. I already had OpenCode agents that could remember project context and route work to specialists for product, marketing, research, and engineering. This week I added Hermes Bots as the agentic CEOs for each company. A Hermes Bot is the persistent operating layer. I g”
“Anvexa AI is hiring — AI Engineer / Forward Deployed Engineer One of the companies at Alan Scott Enterprises, Anvexa AI is building AI agents and automation for real-world enterprise workflows. They’re looking for people who already have some hands-on experience with AI agents, RAG, LLMs and AI automation and want to work on problems beyond just building prototypes. The role sits somewhere between AI Engineering and Forward Deployed Engineering — you’ll build the systems, integrate them with exi”
“One problematic thing about networking is that both parties in the conversation have to spend a considerable amount of time establishing context before they can move on to the actual discussion. Both sides are trying to understand what the other is doing, why, and whether it fits with theirs or is relevant enough that someone should care. This context is usually established through a series of exchanges, or during the first ten minutes of the conversation. Now think that instead of you having to”
“I got tired of my AI agents dying every time I shut my MacBook lid. For the past few months, my entire multi-agent development setup—OpenCode routers, background workers, toolchains, persistent memory—lived strictly on my local machine. It worked great until I needed to step away, put the laptop to sleep, or run a 3-hour autonomous workflow without heating up my lap. So today was migration day. Moved the entire environment over to a dedicated Oracle Cloud ARM64 instance (2 OCPU, 12GB RAM). A few”