AI Hub Monthly · Room or Livestream

Watch Real AI Systems Live. Get the Working Systems After.

Sit ringside in the room or watch the livestream. After the session, get working systems you connect to your AI and stream in.

Real builders open real systems live · 10 room seats total

SATURDAY · OCTOBER 3 · 9:30 AM – 5:30 PM · Estudio Helios, Medellín

Premium room seat if you want ringside. Livestream if you want the view from anywhere. Room and livestream buyers both get the after-session systems.

THE LINEUP

Swipe to see all six. Tap a card to view it full size. On desktop, hover for a larger preview.

03

Can You Build An AI-First Company That Can Keep Printing When You’re Not There? That’s The Fucking Question.

THIS MONTH'S BUILDERS

Six deep dive sessions. One day to model what took them months to engineer.

DALLION KING — AI Teams That Run Your Ventures

01LEAD FEATURED SPEAKER

DALLION KING

Founder & lead architect · Lion's Paw · SigmaVue

AI Teams That Run Your Ventures

Discover the architecture behind organizing AI into operating teams that run your ventures.

WHAT YOUR AI GETS

AGENT TEAM ARCHITECTURE

The session walks through how agents get roles, shared context, tools, and approval gates so a venture keeps moving.

  1. 01Define the roles
  2. 02Share the context
  3. 03Connect the tools
  4. 04Set the approval gates
  5. 05Orchestrate the team

THE SYSTEM

FOR YOUR AI

AGENT TEAM ARCHITECTURE
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

YOUR AI WAITS FOR YOU. THAT'S THE PROBLEM.

Every task still routes through you — open, ask, copy, paste, repeat.

Dallion builds ventures run by teams of AI agents, not one chatbot.

You orchestrate the venture. The team keeps it moving.

Lead featured speaker · AI agent teams that run ventures

MIKE J. ROSENFELD — Build Agents That Think Like Champions — Steal the Playbook

02FEATURED SPEAKER

MIKE J. ROSENFELD

Neuroscientist · Flow State Expert · Trusted optimization expert for Salesforce, Charles Schwab, Novak Djokovic & Jamie Foxx

Build Agents That Think Like Champions — Steal the Playbook

Learn the identity and performance principles behind elite human execution—and how to translate them into clearer roles, context and guardrails for your agents.

WHAT YOUR AI GETS

THE OPERATOR OS

One model, two installs: a personal operating-system worksheet and a copy-paste agent identity template, covered in the session.

  1. 01Identity
  2. 02Perception
  3. 03Feeling
  4. 04Behavior
  5. 05Result

THE SYSTEM

FOR YOUR AI

THE OPERATOR OS
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

YOUR AGENTS INHERIT YOUR OPERATING SYSTEM.

Run the wrong one, and your team — and your AI — run it too.

Neuroscientist · Flow State Expert. Trusted optimization expert for Salesforce, Charles Schwab, Novak Djokovic & Jamie Foxx.

Now the same operating model, installed into your agents.

Featured speaker · the operator OS for humans and agents

JONATHAN TRAN — Why Your Agents Start Work and Never Finish It

03

JONATHAN TRAN

AI builder · Applying Orca to agent workflows

Why Your Agents Start Work and Never Finish It

Give every agent the context to finish the job.

WHAT YOUR AI GETS

PERSISTENT CONTEXT & HANDOFFS

The session covers how to structure work so agents carry the mission and history through to the finish.

  1. 01Set the mission
  2. 02Record what happened
  3. 03Define what comes next
  4. 04Hand off cleanly
  5. 05Finish the job

THE SYSTEM

FOR YOUR AI

PERSISTENT CONTEXT & HANDOFFS
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

AGENTS START WORK. FEW FINISH IT.

The agent forgets. The next one starts from zero. You re-explain the job.

Persistent agents. Shared context. Clean handoffs.

Every agent knows the mission, what happened, and what comes next.

Orca workflows · persistent context & handoffs

ROMAN LEONELLI — He Went All-In on Hermes. Here's What He Wishes He'd Known First.

04

ROMAN LEONELLI

“The Godfather of Hermes” · AI Hub builder · Hermes workflows

He Went All-In on Hermes. Here's What He Wishes He'd Known First.

What worked. What didn't. What to give AI—and what to automate instead.

WHAT YOUR AI GETS

ROMAN'S HONEST LESSONS

The honest version of what worked, what didn't, and what he'd do differently.

  1. 01What worked
  2. 02What didn't
  3. 03What he'd do again
  4. 04What he'd do differently
  5. 05What he'd tell you first

THE SYSTEM

FOR YOUR AI

ROMAN'S HONEST LESSONS
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

YOU CAN LEARN THE HARD WAY. OR GET IN THE ROOM WITH SOMEONE WHO ALREADY DID.

Roman went all-in on Hermes. Now he's sharing the lessons he wishes someone had handed him at the start.

45 minutes. The honest version.

Bring the questions you don't want to learn the answers to the hard way.

Hermes deep dive · the honest lessons

RICHARD REED — Build Apps With AI That Don’t Fall Apart in Production

05

RICHARD REED

Software Engineer at Microsoft

Build Apps With AI That Don’t Fall Apart in Production

Build apps that hold up beyond the demo.

WHAT YOUR AI GETS

PRODUCT-READY APP STACK

The session covers the architecture and defensive controls that take an app from demo to production.

  1. 01Set API boundaries
  2. 02Add mechanical defenses
  3. 03Guard the prompts
  4. 04Plan for attacks
  5. 05Alert on failure

THE SYSTEM

FOR YOUR AI

PRODUCT-READY APP STACK
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

THE DEMO WORKS. THEN REAL USERS SHOW UP.

AI makes it fast to build a prototype. It doesn't make it safe to ship one.

A Microsoft software engineer opens the stack: Quasar, FastAPI, PostgreSQL.

Defenses, guards, and alerting built in — so it survives real users.

Product-ready apps · API defenses & failure visibility

DANIEL GERKE — The AI That Answers Your Clients' Calls

06FEATURED SPEAKER

DANIEL GERKE

Founder · LeadLock · Voice agent platform for agencies

The AI That Answers Your Clients' Calls

Voice agents that answer, book, and follow up — without you picking up the phone.

WHAT YOUR AI GETS

VOICE AGENTS THAT ANSWER

The session walks through how voice agents take calls, book jobs, and write back to the CRM — and the build system that keeps them from breaking.

  1. 01Answer the call
  2. 02Qualify & book
  3. 03Write to CRM
  4. 04Follow up
  5. 05Ship without breaking

THE SYSTEM

FOR YOUR AI

VOICE AGENTS THAT ANSWER
streamed into your AI via MCP

YOU HAND YOUR AI THIS JOB:

“Study this system. Map it to my business. Help me install it.”

It reads the whole system, asks you about your business, and builds your version of this system.

Plus the system map for this build.

YOUR PHONE RINGS. NOBODY ANSWERS.

Every missed call is a missed client — and you can't be on every line at once.

Dan built LeadLock: white-label voice agents that answer, book, and follow up for agencies.

Hear one take a live call. Then open the build system behind it.

Voice agents · white-label platform for agencies

The systems

Watch the system opened live. Stream it into your AI.

You watch real systems opened live in the room. After the session, every build ships as a working MCP module you connect to your AI and stream in. It is not a turnkey build and not a private production repo.

The System

Built live in the room, ready to run.

MCP Module

Connect it to your AI and stream it in.

System Map

How the pieces of the system connect.

Prompts

The instruction patterns used in the walkthrough.

Failure Notes

What broke, and what it cost to find out.

Implementation Notes

Where to start if you adapt it.

Per builder

  • AGENT TEAM ARCHITECTURE · streamed via MCP
  • THE OPERATOR OS · streamed via MCP
  • PERSISTENT CONTEXT & HANDOFFS · streamed via MCP
  • ROMAN'S HONEST LESSONS · streamed via MCP
  • PRODUCT-READY APP STACK · streamed via MCP
  • VOICE AGENTS THAT ANSWER · streamed via MCP

Stop guessing from finished screenshots.

Watch the working systems get opened live.

Most of what you see online is a finished screenshot: the result, never the structure behind it. In this session, builders open working AI systems live and walk through how they are put together, what broke, and what they changed. You can watch ringside in the room or from the livestream.

This is not slides and it is not a recorded webinar. It is real systems on screen, questioned live, with the reasoning behind the choices said out loud, including the approaches that were rejected and why.

After the session, every system built live ships as a working MCP module. You connect it to your AI, stream it in, and the machine keeps running long after the room clears out.

  1. 01

    The builder opens the live system

    The structure, the prompting approach, the routing: walked through live, end to end, on screen.

  2. 02

    The room pulls it apart

    The room questions the design until the failure modes and the fixes are on the table.

  3. 03

    The session becomes working systems

    Afterwards every system from the session is shipped as a connectable module. Plug it into your AI via MCP, stream it in, and it goes to work in your stack the same night.

What it comes down to

Live
Working systems opened and explained on screen
One session
Real production systems, opened and explained, in one focused working session
Systems
Working MCP modules you connect to your AI and stream in
Paid once
Room seat or livestream access · one payment

Seeing how something is actually structured is worth more than another explanation of it. Which is why the room is built the way it is.

What The Session Leaves You With

After the session, every system built live ships as a module you connect to your AI and stream in.

Sample system

03-ai-employee · System Notes
# AI EMPLOYEE · SYSTEM MODULE

## WHAT THIS SYSTEM IS
The live system from the walkthrough, ready to connect to your AI.
It is not a turnkey build and not a private production repo.

## WHAT TO CONNECT
1. What work enters the system.
2. How the operator reviews output.
3. Where the failure modes showed up.
4. What your AI/team should map before implementation.

## ADAPTATION CHECK
- What part fits your business?
- What needs human approval?
- What should not be automated yet?

The systems you connect

ai-hub-systems/ / 01-long-running-agents/ / Agent Loop / Watchdog / 02-local-stack/ / Hardware Sizing / Deployment / 03-ai-employee/ / System Notes / Intake SOP / 04-agent-fleet/ / Routing Map / Failure Modes

Working systems. Connect them to your AI, stream them in, and build on top.

The protocol

A small room. A live feed. One system on screen.

One table with ringside room seats, a livestream for everyone watching from elsewhere, and one working system on screen at a time. Everyone gets the working systems after.

Ringside room seats

The room is kept small so every seat is inside the conversation and you can interrupt a builder mid-demo.

Livestream access

If you are not in Medellín, you watch the same systems opened live from anywhere, and get the same systems after.

Systems on screen

The screen carries the working system: cameras stay on what is running, not on an audience.

The systems after

Once the day is organized, the systems ship to everyone who watched — room or livestream — ready to connect to your AI and stream in.

Room seats are limited. Livestream access stays open until the session starts.

The Agenda

  1. 9:30Arrival & coffee
  2. 10:00Deep dives · Block I
  3. 1:30Lunch
  4. 2:15Deep dives · Block II
  5. 5:15Connections
  6. 5:30End

Saturday, October 3, 2026 · Medellín studio boardroom. Lineup and running order can shift as builders and sessions are confirmed.

CHOOSE HOW YOU WANT TO WATCH.

Take a ringside room seat in Medellín, or watch the same systems opened live on the stream from anywhere. Both include the working systems after the session.

CHOOSE YOUR ACCESS

Room seats are limited. Lineup may shift; the format does not.

What the access buys back.

This is not another explanation of AI. It is a live look at how working systems are structured, and systems that help your AI or team map what to do after.

Watching the systems opened live

1 day

Time invested

Builders open working AI systems live, start to finish.

Working systems

What gets delivered

Every build shipped as an MCP module you connect to your AI and stream in.

Paid once

Cost

One payment. Room seat or livestream access.

Room or live

Access

Ringside in Medellín, or watching the same feed from anywhere.

Working from finished screenshots

Structure hidden

Guesswork

You see the result, never how it was put together.

Unanswered

No one to ask

No way to question a choice while the system is in front of you.

Unknown

Rejected paths

You repeat approaches other builders already abandoned.

Nothing organized

After the fact

Nothing structured for your AI or team to work from.

You are buying a live look at how working systems are structured, and something organized to work from afterwards.

CHOOSE YOUR ACCESS→

Room seats are limited. Lineup may shift; the format does not.

FAQs

Room seat or livestream access

CHOOSE YOUR ACCESS

Founding Edition access. One payment. Room seat in Medellín or livestream access from anywhere.