AI Marketing Team Builder vs Durable Sessions UX Framework
// TL;DR
These two skills solve completely different problems and do not compete. If you are a marketer building an autonomous AI team inside Claude Code, use Grace Leung's AI Marketing Team Builder. If you are an engineer fixing broken AI chat streaming, disconnections, and multi-device UX, use Christensen's Durable Sessions Framework. Pick based on your role: marketer or infrastructure engineer. There is no overlap — using one does not eliminate the need for the other.
// HOW DO THEY COMPARE?
| Dimension | Grace Leung AI Marketing Team Builder | Christensen Durable Sessions AI UX Framework |
|---|---|---|
| Best for | Marketers systematising recurring campaigns into an AI agent team | Engineers diagnosing and fixing broken AI chat streaming and real-time UX |
| Problem domain | Marketing workflow automation and multi-agent orchestration | Streaming architecture resilience, multi-device continuity, and live agent control |
| Complexity | Moderate — requires brand assets, folder setup, and iterative agent building across 14 steps | High — requires deep understanding of streaming protocols, pub/sub, WebSockets, and distributed systems |
| Time to apply | A few hours to build your first agents; days to build a full team with routing rules | Days to weeks for a full architectural migration from SSE to Durable Sessions |
| Prerequisites | Claude Code access, brand context files, branded templates, marketing function map | Existing AI product with streaming, knowledge of SSE/WebSockets, backend engineering capability |
| Output type | Campaign deliverables: decks, social creatives, landing pages, research briefs, ad copy | Redesigned streaming architecture with resilient delivery, cross-surface continuity, and live control |
| Target user | Marketers and marketing ops teams | Backend engineers and AI product architects |
| Tool dependency | Claude Code, optional Notion and MCP integrations | Pub/sub infrastructure (e.g. Ably), WebSocket transport, session state management layer |
| Creator background | Grace Leung — marketing-focused Claude Code educator | Mike Christensen at Ably — real-time infrastructure and AI UX specialist |
| Multi-agent relevance | Builds and orchestrates multiple marketing agents with routing rules | Solves the infrastructure problem of delivering multi-agent output to clients without bottlenecks |
What does the Grace Leung AI Marketing Team Builder do?
The AI Marketing Team Builder is a step-by-step framework for constructing a fully autonomous marketing team inside Claude Code. It follows a strict four-layer sequence: map your marketing functions, convert each repeatable task into a reusable skill, group skills into focused agent roles, then connect agents as a coordinated team.
The framework emphasises pre-loading brand context (voice guides, style guides, strategy docs) before building anything. It introduces the Reference-Based Method — where Claude analyses an existing branded template before generating a skill — to anchor output quality to real brand standards rather than generic prompts. The end result is a system where you can issue a single campaign brief and receive a full package of research, strategy, social posts, landing pages, and creatives in roughly ten minutes.
Key capabilities include dedicated non-overlapping agent roles (Data Analyst, Content Creator, Market Researcher, Creative Designer, Campaign Strategist), a shared Notion task board for human-AI collaboration, and mobile remote control for dispatching tasks from anywhere.
What does the Christensen Durable Sessions AI UX Framework do?
The Durable Sessions Framework diagnoses why AI chat products break under real-world conditions and provides an architectural pattern to fix them. It identifies the Single-Connection Trap — where stream health is coupled to one client's connection — as the root cause of most AI UX failures.
The framework prescribes three foundational capabilities every production AI product needs: Resilient Delivery (streams survive disconnections), Continuity Across Surfaces (sessions follow users across tabs and devices), and Live Control (clients can steer or cancel agent work mid-generation). The solution is a Durable Sessions layer — a persistent, shared pub/sub channel between agents and clients that decouples the two entirely.
This is deep infrastructure work. It addresses the SSE Resume-Cancel Conflict (closing a connection is ambiguous between disconnect and cancel), the Orchestrator Dual-Purpose Problem (orchestrators forced to relay sub-agent updates), and the need for bidirectional transport when live control is required.
How do they compare?
These two skills operate in entirely different domains and solve fundamentally different problems. Comparing them on a single axis is misleading, so it is important to be precise about what each one does and does not do.
The AI Marketing Team Builder is a workflow and agent design framework for non-technical marketers. It lives inside Claude Code, produces marketing deliverables, and assumes the underlying infrastructure (Claude's runtime, file system, API connections) already works. It never addresses how streaming tokens reach a user or what happens when a connection drops.
The Durable Sessions Framework is a streaming architecture pattern for engineers building AI-powered products. It does not produce any marketing content. It does not build agents in the marketing sense. It fixes the plumbing that makes AI product experiences resilient, multi-device, and controllable. It is relevant to any team shipping an AI chat or agent product to real users, regardless of domain.
The one area of conceptual overlap is multi-agent orchestration. Both frameworks deal with multiple agents, but at different layers. The Marketing Team Builder defines what agents do and when to route tasks between them. The Durable Sessions Framework solves how multiple agents' outputs reach clients without an orchestrator bottleneck. In a production system, you could conceivably need both — one for agent logic, one for delivery infrastructure — but in practice their audiences rarely overlap.
Which should you choose?
Choose the AI Marketing Team Builder if you are a marketer or marketing team lead who wants to move beyond one-off Claude prompts into a persistent, structured agent system that produces campaign deliverables on-brand and at speed. You need Claude Code access and brand assets, not engineering expertise.
Choose the Durable Sessions Framework if you are an engineer or product architect building an AI-powered product and your users experience broken streams, lost responses on mobile, no multi-device sync, or an unreliable stop button. You need backend infrastructure skills and an understanding of streaming protocols.
If you are a technical founder building an AI marketing product for others to use, you might need both: the marketing team builder's agent design principles for the product logic layer, and Durable Sessions for the delivery infrastructure layer. But for most people, the choice is obvious from your job title alone.
Can these two frameworks work together?
Yes, but only in a narrow scenario. If you were building a SaaS product that lets users run AI marketing agents in a browser, you would use the Marketing Team Builder's principles to design the agent layer and the Durable Sessions pattern to ensure those agents' outputs survive disconnections and work across devices. For the vast majority of users — individual marketers or product engineers — you will use one or the other, never both.
// FREQUENTLY ASKED QUESTIONS
Is the AI Marketing Team Builder or Durable Sessions Framework better for building AI agents?
They handle different layers of agent systems. The Marketing Team Builder defines agent roles, skills, and routing logic for marketing tasks inside Claude Code. The Durable Sessions Framework solves how agent output reaches clients reliably across devices and disconnections. One is agent design; the other is agent delivery infrastructure.
Do I need coding skills to use the Grace Leung AI Marketing Team Builder?
No. The framework is designed for marketers using Claude Code's interface. You need brand assets (voice guide, templates, strategy docs) and familiarity with Claude Code commands like /agents and /plugin, but no programming knowledge. The Durable Sessions Framework, by contrast, requires backend engineering skills.
Can the Durable Sessions Framework help me create marketing content?
No. The Durable Sessions Framework is a streaming architecture pattern for engineers. It solves connection resilience, multi-device continuity, and live agent control. It produces no content of any kind. For marketing content automation, use the AI Marketing Team Builder instead.
What is the main problem the Durable Sessions Framework solves?
It solves the Single-Connection Trap — where your AI product's response stream breaks when a user's connection drops, switches devices, or tries to cancel generation. It introduces a persistent session layer between agents and clients so streams survive disconnections, work across devices, and support live user control.
How long does it take to set up a full AI marketing team with Claude Code?
Expect a few hours to build your first agents and skills, and several days to construct a full five-agent team with routing rules, style libraries, and a connected task board. The Reference-Based Method for each skill adds time upfront but dramatically improves output quality on first run.
Should I use both frameworks together?
Only if you are building a SaaS product that serves AI marketing agents to users through a web or mobile interface. In that case, use the Marketing Team Builder's principles for agent logic and the Durable Sessions pattern for reliable delivery. Most individuals will use one or the other based on whether they are a marketer or an engineer.
What is the Reference-Based Method in the AI Marketing Team Builder?
It is a skill-building approach where you give Claude an existing branded template, have it generate a detailed analysis report of that template's patterns, then build the skill from the analysis. This anchors output to real brand standards rather than starting from a blank prompt, producing significantly higher-fidelity deliverables.
Why does the Durable Sessions Framework say SSE is not enough for AI products?
SSE is one-way — the server streams to the client, but the client cannot send messages back. This means closing a connection is ambiguous: it could mean the user disconnected or pressed stop. Resume and cancel become mutually exclusive. Bidirectional transport like WebSockets is needed for live control features.