Agentics Foundation - Seattle Chapter - Virtual Roundtable
This is an open-door community meetup open to members, friends, and guests. Come hang out with builders, tinkerers, and AI-curious folks at our public online event on Zoom with video and live interactive discussions.
What is this Event about?
This session is all about community: who's in the room, what they're building, and where agentic AI is actually headed.
Who should join?
If you are learning or practicing Agentic AI and are in the Greater Seattle Area, this is your opportunity to network and meet with AI experts.
What to expect?
Get insights into breakthroughs in AI, computer science, and advanced applications of AI at the cutting edge of technology.
Seattle Chapter Intros — quick round of who we are and what we're working on
<<<< Lightning Presentations >>>>
We have three 10 minutes presentations about pressing topics and cool things our members are building. This month you will hear about:
### Robert E. Lee, Claude-Telegram-Mirror (CTM) -
https://www.linkedin.com/in/robertgpt/
https://github.com/robertelee78/claude-telegram-mirror:
The Most Fragile Line of Code: Controlling an AI Agent From Your Phone
What does it take to drive Claude Code from a Telegram thread on your phone (approve a rm -rf, steer the agent as it works) when it only speaks through a terminal? This talk follows claude-telegram-mirror, a ~19k-line Rust bridge, and zooms in on its hardest problem: getting a human's answer back into an interactive CLI. The first fix simulated keystrokes into a tmux pane, a ~4.7-second dance that raced the terminal and broke constantly. The real fix was to stop typing and speak the agent's own protocol: a blocking lifecycle hook that hands the answer back as structured data, cutting delivery from ~4.7 seconds to microseconds. It's also a case study in building with AI, and shipping it without trusting it blindly.
### Vincent Kovar, Agentic Foundation (Seattle), AI Tinkerers (Lynwood)- https://www.linkedin.com/in/vincentkovar
The 78%: Why AI Education Is Failing Humans
After earning around 90 badges and certifications in AI as well as conducting a 100-day study of 80+ seminars, I’ve found that current pedagogies (at best) only serve 22% of the adult population, effectively marginalizing 78% of potential adopters. It is easy to blame poor content and “salsey” product pitching. It is even easier to blame aspiring learners. But we don’t build the foundations of artificial Intelligence with random YouTube videos, confusing GitHub dumps, or by blindly clicking through Jupyter Notebooks. And, ~80% of the time, those won’t build human intelligence either. We need to shift how we think about skill acquisition and apply the foundational principles of machine learning (scaffolding, gradient-based feedback, and weight adjustment) to human instruction. Let’s explore how.
### Andy Zhulenev, Founder OneForce.com
TOPIC: NextGen Productivity With AI
https://www.linkedin.com/in/andreyzhulenev/
DESCRIPTION: The solution we are building redefines personal productivity for people working with AI. The user doesn’t need to learn AI. The user can simply chat with AI as their personal assistant. The system would track all tasks, projects, contacts, calendars, and emails, and manage folders and files, without requiring the user to log in to their software tools.
Whether you're shipping production AI systems or just ran your first LLM API call last week, this is your room.
What to bring?
Bring questions you've been sitting on, and something you're genuinely curious about.
About Us