Organization of the Future: Simulating the AI-Native Teams Workshop

#

### Hosted by Ataccama // Karlín, Prague 8

***

Agent-Driven Meetup Prague: Build Like a POD — Simulating the AI-Native Development Team

***

## Short Description

Your org chart is killing your agents. Join us to run a live simulation of POD-based teams making decisions with AI agents — not talking about it.

***

## Full Description

***

### This Is Agent-Driven Meetup Prague.

If you've been here before, you know what this is: no bullshit theory, no vendor pitches, no "AI is going to change everything" presentations. We build things, we break things, we figure out what actually works when you put agents into production in real organizations.
This edition goes one level up from the code. Not how to build agents — how to build the teams that operate them.
Because here's the thing nobody is saying out loud at these meetups: the agents are fine. The organizational structure around them is broken.

***

### The Real Problem We Keep Not Solving

Every team at every meetup we've run is using agents the same way. One developer has a Claude window open. Another has Cursor. Someone on Slack posted a prompt they found useful. There's no shared context, no shared role assignment, no way for the agents to be collectively accountable to the team's actual goals.
You're running 21st-century AI on a 20th-century org chart.
The org chart was designed for predictable, repeatable, decomposable work — industrial-era assumptions baked into every hierarchy, every feature team, every sprint board. It doesn't know what to do with agents because agents aren't tasks. They're capable entities. They need roles. They need context. They need to fit into a structure that was designed to hold them.
Dave Gray figured this out in 2012, before agents existed.

He called it the Connected Company. The core unit is a POD — a small, autonomous, cross-functional team with everything it needs to serve its customer independently, connecting to other PODs through explicit interfaces rather than management layers.

***

### What Actually Happens at This Meetup

You walk in. You get assigned to a POD.
Each POD is a cross-functional unit inside NovaTech — a fictional (but uncomfortably familiar) Czech B2B SaaS company that is mid-transformation from a traditional hierarchy into a connected, POD-based structure. Product PODs. Revenue PODs. Infrastructure PODs. Customer Success PODs. Each with a challenge scoped to their function.
The company faces one decision: should NovaTech launch in Germany in Q3?
No POD has the full picture. The Product POD knows what the roadmap costs. The Revenue POD has deals on the table. The Infrastructure POD has a data residency timeline. The Customer Success POD knows what they can't support. The decision is genuinely hard. The PODs have to figure it out together — without a manager breaking the tie.
Your POD gets an AI agent. Not a tool. A team member with a role you define, a name you give it, a brief you write. Research Analyst. Technical Due Diligence Analyst. Revenue Strategist. You decide. You brief it. You work with what it gives you.
You have 45 minutes. Then all PODs assemble for the Company Council — the Connected Company's answer to a board meeting, except there's no board. Just PODs, with their recommendations, their conflicts, and the requirement to reach a collective position without anyone having authority to overrule anyone else.
That part is the hard part. It's supposed to be.
Afterwards, we debrief. What did the POD structure change about how you worked? What did you do with your agent that you hadn't done before? Where did the Company Council break down — and why? These aren't hypothetical questions after this exercise. They're questions you just lived the answer to.

***

### Why This Is the Right Conversation for This Community

We've spent the last several meetups on the technical layer: orchestration, multi-agent coordination, memory and context management, governance patterns. We've built real things and learned real things.
But the people at these meetups go back to teams that are still organized the same way they were before agents existed. The technical knowledge doesn't transfer because there's nowhere for it to land. You know how to build a multi-agent pipeline. Your team has no structure to deploy it into.
POD architecture is the landing pad.
When your team is a POD — with a defined customer, a defined purpose, local decision-making authority, and explicit interfaces to other PODs — the question "where does the agent fit?" has an actual answer. The agent has a role. The role has a brief. The brief has a deliverable. The deliverable has a place in the POD's output to its customers. This is not complicated. But it requires the organizational structure to exist first.
This meetup builds that structure — in a simulation — so you can feel what it's like before you try to convince your organization to actually do it.

***

### What You'll Take Away

• Direct, embodied experience of what POD autonomy feels like versus feature-team execution — the difference is not subtle
• A working model for how to assign AI agents functional roles within a team structure, not just task roles within a conversation
• Concrete intuition for where inter-POD coordination fails and why that failure mode is the hardest problem in connected organization design
• The specific insight that agent architecture and organization architecture are the same problem — you can't solve one without the other

***

### Who This Is For

Everyone who has been to Agent-Driven Meetup before and wants to understand why the agent knowledge isn't transferring into their actual team.
Everyone who hasn't been to Agent-Driven Meetup before but works on a development team that is trying to figure out how to actually integrate AI into how they work — not just into individual workflows.
Tech leads. Engineering managers. Developers with opinions about team structure. CTOs at companies that are rethinking how they organize. Product people who want to understand why their developers relate to agents so individually when the problem is collective.
You do not need to have read The Connected Company. You need to be willing to be uncomfortable in a simulation with strangers for 45 minutes. That's it.

***

### Schedule

| Time | |
| ---- | --- |
| 18:00 | Doors, drinks, talking to people you don't know yet |
| 18:30 | Opening frame: POD architecture in 20 minutes, agents as team members |
| 18:50 | POD assignment, briefing, agent naming |
| 19:00 | POD Work Session — 45 minutes live |
| 19:45 | Company Council — all PODs, one decision, no manager |
| 20:15 | Structured debrief |
| 20:45 | Open floor, Q&A, networking |
| 21:30 | End |

***

### Location

Ataccama Sokolovská 685/136f, Karlín, Prague 8
Metro: Invalidovna (line B), 5 min walk. Tram: Urxova.
Ataccama builds enterprise data trust platforms and has been doing it from Karlín since 2007. They know what it means to scale a software organization. We're grateful they're hosting.

***

### Capacity

40 people maximum. This is not a meetup you can attend passively. The simulation requires real PODs of 5–7 people. Above 40, the Company Council stops working. Register early or don't register.

***

### Bring

• A laptop (one per POD minimum, two is better)
• Your actual opinions about how teams should be organized — you will use them
• Optional and appreciated: skim The Connected Company by Dave Gray before you come. The summary is enough.

***

Agent-Driven Meetup Prague is the leading technical AI community in Central Europe focused on production-grade agent systems, organizational AI integration, and the engineering disciplines required to govern both. We meet, we build, we figure out what actually works.
Organizer: Jakub Bareš — [bares.jakub@gmail.com](mailto:bares.jakub@gmail.com)