Generated concept art: she laces a gap in the street shut with red rope
Devlog 3 · Epic AI MegaJam · Coordination

Coordinating people and AI agents Our communication setup during the jam

Sandscape team · November 14, 2026 · 3 min read

We started False Depth two days ago, and the first playable must be done by tomorrow night. For this third post, we're looking at how we keep three people and many AI agents working in the same direction, including the communication setup we've already had to rethink.

More work happening at once

AI is speeding up development a lot. Sandscape is at the core: we set the art direction there, generate assets that follow it, and use its skills to drive the Unreal editor. With coding agents outside Sandscape, we still have binary files that can't be merged and people needing separate work. Each of us starts another task while an agent works, so there are many agents to coordinate. Some of us work day and night; others have work or family alongside the jam. People come and go while tasks keep running.

Three people each run several AI agents at once, and every agent works on files in the same Unreal project. Two agents reach for the same map file, which cannot be merged. Each agent works on its own copy and pushes changes here OUR GIT REPO ON AZURE DEVOPS SEB LAURENT NAT AI AGENTS .uasset.uasset.uasset.umap.uasset.uasset .uasset.umap.uasset.uasset.uasset .uasset.uasset.uasset.uasset.umap.uasset .uasset.uasset.uasset.uasset.uasset.uasset .umap Two agents edit the same map; binary files cannot be merged

Our first message board

From day 0, we had a producer agent overseeing the project, keeping the task board and managing information. It flagged problems early, including overlapping file edits and stale tasks. For messages between machines, we made a quick hack: a small repository where agents could leave messages. It was useful, but agents picked messages up themselves. We lost track of information, and some landed in chats that didn't need it. Meanwhile, we were talking on Teams, disconnected from all of this.

How our setup changed during the jam: a producer and a message repo on day 0, Discord and a desk on every machine on day 2, tagged questions after that, and next, bringing it into Sandscape. Two roadblocks are marked. DAY 0 DAY 2 NEXT PRODUCER Tracks the plan and flags problems MESSAGE REPO Agents leave messages for other machines People lost track of messages; some reached unrelated chats while we used Teams separately DISCORD People and agents read and post here A DESK PER MACHINE Handles local questions The desk tried to follow every chat and explained fragments without enough context TAGGED QUESTIONS Replies return only to the asking chat IN SANDSCAPE Being considered
Each change in the setup came from a communication problem we hit.

A producer and local desks

On day 2, we moved to Discord, where we and our agents can read and post. One channel holds decisions for people; another is for machines. The producer could see and flag tasks on its own machine, which suggested a local producer for everyone: a desk. One central producer oversees the project and follows up until problems are handled. Each desk picks up local questions, talks to the other desks and passes information into the right conversation. Often, we don't need to be there: a question can reach the agent that knows the answer through the desks. Sessions start with a large fixed load of instructions and tools and grow over days; we compact the producer's and desks' context to keep idle agents cheap to wake. A watcher script watches Discord, the task board and repositories instead of agents checking on a timer: watching costs nothing, and only a real change wakes an agent and costs a turn.

A question travels from the level design chat on Seb's machine to Discord, then to Laurent's desk, which hands it to the chat working on Clara's captured clips. The answer travels back through Discord and Seb's desk to the chat that asked, and nowhere else. The producer watches the channel. # DISCORD #board for messages between machines #decisions for decisions that need a person PRODUCER Routes and follows up SEB'S MACHINELAURENT'S MACHINENAT'S MACHINE Desk Desk Desk Level design Bent street Controller fix Clara's face Capturedclips Rooftop scene Level art Red: Seb's level design chat asks; Laurent's desk delivers to "captured clips" Gold: reply via Seb's desk to the asking chat
The desks carry a question across machines and return the answer to its conversation.

Fixing the desk

The first desk tried to explain every conversation on its machine, using fragments without their context. Its summaries confused us, answered questions resurfaced, and its list went stale. Using it was like handling the chats ourselves, only more complicated. We stopped it watching. Now an agent posts a tagged question naming who asked, which conversation it came from and who should answer. The producer routes questions with no known destination and chases unanswered ones. Desks deliver them; answers return only to the asking conversation. The agent with the context should explain, while the others carry its message.

What we want to keep

We think we'll keep using this setup whenever we build in Unreal. We're also thinking about bringing some of these workflows directly into Sandscape, so working with Unreal gets easier for everyone. Tomorrow's last post will return to False Depth and the playable we need to finish.

The seriesOur jam devlogs

Play False Depth on itch.io