# Context as Code, DevX as Leverage: Accelerating Software with Multi‑Agent Workflows

Page: https://stenobird.com/podcast/ai-engineering-podcast/context-as-code-devx-as-leverage-accelerating-software-with-multi-agent-workflows
Text version: https://stenobird.com/podcast/ai-engineering-podcast/context-as-code-devx-as-leverage-accelerating-software-with-multi-agent-workflows.md
Podcast: [AI Engineering Podcast](https://stenobird.com/podcast/ai-engineering-podcast)
Published: 2025-11-24T01:02:17+00:00
Episode link: https://www.aiengineeringpodcast.com/agor-multi-player-multi-agent-engineering-episode-70
Audio file: https://op3.dev/e/dts.podtrac.com/redirect.mp3/serve.podhome.fm/episode/f6ff0caa-931b-4c08-bfdd-08dc7f5cd336/6389954251734695544a901724-7644-4990-9b6d-00f12ef3878e.mp3
Processing state: processed
JSON: https://stenobird.com/v1/public/podcasts/ai-engineering-podcast/episodes/context-as-code-devx-as-leverage-accelerating-software-with-multi-agent-workflows
Duration seconds: 3589

## Resource

Software engineering is shifting from manual coding to the orchestration of multi-agent workflows. Max Beauchemin introduces Agor, an open-source platform designed to manage these agents through a spatial, multiplayer canvas.

## Highlights
- Main idea: The developer's role is evolving from writing lines of code to managing a fleet of specialized AI agents
- Practical takeaway: Using Git worktrees and spatial canvases can help developers manage multiple concurrent AI-driven features without context loss
- Failure mode: Unstructured context and long-running sessions can lead to context window explosion and agent hallucination
- Technical innovation: Agor utilizes the Model Context Protocol (MCP) to allow agents to interact with the system and each other
- Efficiency gain: Multi-agent parallelization allows a parent agent to spawn sub-sessions for specific tasks, like large-scale refactoring, and aggregate results

## Topics

AI Engineering, Multi-Agent Systems, Software Orchestration, Model Context Protocol, Developer Experience, Git Worktrees, Open Source, Agor

## Chapters
- 5:35 — The AI-First Developer Reflex: Max discusses adopting an 'AI-first' approach for nearly all tasks, from coding to legal research and blog writing.
- 10:00 — Shifting Bottlenecks in AI Workflows: As execution speeds up, the new bottlenecks move to code review, QA, and managing the increasing complexity of context.
- 24:10 — Managing Complexity with Git Worktrees: How to use Git worktrees to effectively manage multiple simultaneous feature branches and AI-driven tasks.
- 28:20 — Introducing Agor: Multi-Agent Orchestration: An overview of Agor's spatial canvas, featuring multiplayer cursors, templated prompts, and session management.
- 32:55 — Automated Environment Management: Using Agor to automate Docker Compose and port management, ensuring seamless shared development environments.
- 46:20 — Agent-to-Agent Orchestration: Deep dive into complex workflows where agents spawn sub-sessions to perform parallelized tasks like JS to TS migrations.
- 54:50 — Session Forking and Context Control: The benefits of forking sessions to experiment with new prompts without polluting the main development context.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/ai-engineering-podcast/episodes/context-as-code-devx-as-leverage-accelerating-software-with-multi-agent-workflows/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/ai-engineering-podcast/context-as-code-devx-as-leverage-accelerating-software-with-multi-agent-workflows.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.