​Online workshop for AI-native builders.

​No slides. No theory. Live demos.

​Founders and operators share the actual workflows running their companies — agents, harnesses, knowledge graphs, personal AI OSes, skills, evals — in real time.

​What it is

​A 120-minute live workshop. Each operator gets ~15 minutes to demo one end-to-end AI-native workflow they run today, followed by ~10 minutes of moderated Q&A. This is session four of a monthly series.

​Who it's for

​Founders and operators rebuilding their companies around AI. Especially founders, business leaders, and the people inside companies learning AI-native best practices — engineering, ops, growth, marketing, sales. If you're shipping agents in production, this is for you.

​What's inside

• ​~15-min screen-share demos of real, in-production workflows

• ​Moderated Q&A after each — the only way to ask the speakers live is to be here

• ​Recording + written case studies sent to all registrants

​Why attend live

​The recording will be heavily edited and shipped as case studies. The unedited live Q&A only happens once. First invites to the curated cybos.ai operator community will go out to the live audience.

​Speakers

​Pierson Marks — Founder & CEO, Jellypod

​Demo: "The software factory: from user issues to validated PRs while you sleep"

​Pierson runs Jellypod, an AI podcast platform built by a two-person team. He'll demo his human-in-the-loop "software factory": every morning a cloud agent reads PostHog session replays, Vercel logs, and Intercom messages against the roadmap, then files and de-duplicates issues in Linear. From there, Claude agents pick up bugs in parallel — writing fixes, opening PRs, spinning up preview branches, and validating each fix with a Playwright browser before attaching a screen recording. All live: the prompt, the MCP connections, the routines config, and a real run from scheduled task to validated PR.

​X · LinkedIn · https://jellypod.ai

​Pavel Muntyan — Founder, mf0.ai

​Demo: "A graph-memory brain for your AI"

​Pavel builds mf0.e, an AI aggregator built around a graph-based persistent memory. It's a "digital brain" you can actually see and optimize, with two hemispheres and nodes branching from "I" and "Interests." Search prioritizes your most-connected memories and stays on-topic to avoid drift. He'll demo it live: one shared context across ChatGPT, Gemini, and other models, "AI Opinion" (every model answers the same question in parallel), and "Fusion" (five models answer, one fact-checks, and you get a structured document).

​X · https://mf0.ai

​Zecheng Zhang — Founder, Strukto (creator of Mirage)

​Demo: "Mirage: one clean environment for your agents"

​Zecheng builds Strukto and its open-source project Mirage. His thesis: the bottleneck for agents is increasingly the environment, not the model. Mirage turns scattered data and services (files, email, Slack, Notion) into a single virtual file system that an agent drives from one terminal, cutting token cost, improving performance, and giving you fine-grained control over what each agent can see and do. He'll demo Mirage live: agents working across a unified environment, with governance over every channel.

​X · https://strukto.ai · Mirage (open source)

​Host: Stepan Gershuni (LinkedIn | X)

​About cyber•Fund & the Monastery

​cyber•Fund backs AI-native founders. The Monastery is our 12-week accelerator for teams building AI-native from day one — $2M uncapped SAFE, 8-team cohort, SF kickoff and a demo day in October. Tooling, GPU credits, coaching, deep-focus monastic environment. We'll share how to apply at the end of the session.