Durable Sessions AI UX vs Science-Ranked To-Do List
// TL;DR
These two frameworks solve completely different problems and do not compete. If you are building or auditing an AI chat product with streaming, disconnection, or multi-device issues, use the Christensen Durable Sessions AI UX Framework. If you are struggling with personal or professional task completion, procrastination, or an overwhelming to-do list, use Dr. Jamie's Science-Ranked To-Do List System. Pick the one that matches your problem domain — there is zero overlap.
// HOW DO THEY COMPARE?
| Dimension | Christensen Durable Sessions AI UX Framework | Dr. Jamie's Science-Ranked To-Do List System |
|---|---|---|
| Best for | Engineering teams building AI-powered chat or agent products that need resilient, multi-device streaming | Individuals or knowledge workers who want to complete more tasks with less procrastination |
| Domain | AI product architecture and real-time infrastructure | Personal and professional productivity / behavioral science |
| Complexity | High — requires understanding of streaming protocols (SSE, WebSockets), pub/sub systems, and agent topologies | Low to moderate — requires self-reflection and consistent daily practice, no technical prerequisites |
| Time to apply | Days to weeks — involves architectural redesign, infrastructure changes, and testing across failure modes | Hours to days — can start with a brain dump and values audit today, full system in place within a week |
| Prerequisites | Existing AI product with streaming architecture, knowledge of SSE/WebSockets, access to infrastructure (e.g., pub/sub layer) | A list of tasks and willingness to follow a structured workflow; pen and paper or any task app |
| Output type | Architectural redesign — a durable sessions layer, transport migration plan, validated resilience/control/continuity | A prioritized, time-blocked daily plan with implementation intentions and habitualized recurring tasks |
| Evidence base | Engineering best practices and real-world failure mode analysis from production AI products | Peer-reviewed behavioral science including a 94-study meta-analysis on implementation intentions |
| Creator background | Mike Christensen (Ably) — real-time infrastructure and AI UX, presented at AI Engineer conference | Dr. Jamie (via Ali Abdaal) — behavioral science researcher applying peer-reviewed findings to productivity |
| Who should NOT use this | Anyone without an AI product — this is irrelevant to personal productivity or task management | Anyone looking to fix AI product streaming issues — this framework does not address software architecture |
| Ongoing maintenance | Operational monitoring of session layer, transport health, and multi-agent pub/sub channels | Daily top-three selection, weekly values audit, ongoing habit migration of recurring tasks |
What does the Christensen Durable Sessions AI UX Framework do?
The Christensen Durable Sessions AI UX Framework diagnoses and fixes the infrastructure layer that most AI chat products get wrong. It was introduced by Mike Christensen of Ably at the AI Engineer conference and targets a specific failure: the Single-Connection Trap, where a user's AI response stream dies the moment their network hiccups, their tab closes, or they switch devices.
The framework identifies 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 (users can steer or stop an agent mid-generation). It then prescribes an architectural pattern called Durable Sessions — a persistent, shared pub/sub layer that sits between agents and clients, decoupling the two so neither depends on the other's connection state.
The workflow walks engineering teams through auditing their current SSE or WebSocket setup, identifying failure modes like the SSE Resume-Cancel Conflict and the Orchestrator Dual-Purpose Problem, then redesigning around durable sessions. It is a deep infrastructure framework that requires real engineering effort but solves problems that no amount of prompt engineering or model improvement can touch.
What does Dr. Jamie's Science-Ranked To-Do List System do?
Dr. Jamie's Science-Ranked To-Do List System is a personal productivity framework that ranks every major to-do list strategy using peer-reviewed behavioral science. Popularized through Ali Abdaal's channel, it gives each method a tier rating (S through F) based on research evidence for actually completing tasks.
The system's two S-tier strategies are Implementation Intentions (specifying when, where, and how you will do a task, backed by a 94-study meta-analysis) and Specific Time Blocking (assigning individual tasks to exact calendar slots, which functions as an implementation intention). A-tier strategies include Eat the Frog (hardest task first), the Eisenhower Decision Matrix, Self-Concordant Tasks (values-aligned lists), and Monotasking. F-tier — and this is emphatic — is multitasking, which fails in every study.
The workflow is sequential: brain dump everything, run a values audit to prune the list, apply the Eisenhower matrix, select a daily top three, sequence with the hardest task first, attach implementation intentions, time-block on your calendar, habitualize recurring items, and eliminate multitasking. It is designed for immediate, same-day application.
How do they compare?
They do not compete. These frameworks operate in entirely different domains with zero functional overlap.
The Durable Sessions framework is for software engineering teams building AI products. Its inputs are streaming architectures, agent topologies, and client surfaces. Its output is an infrastructure redesign. You need production systems, engineering resources, and familiarity with real-time protocols to use it.
The Science-Ranked To-Do List System is for individuals managing their own task completion. Its inputs are a task list, a primary pain point, and optionally your chronotype. Its output is a structured daily plan. You need a pen and five minutes to start.
The only conceptual thread they share is a concern with systems thinking — both argue that your current default approach has structural flaws that no amount of effort will overcome. Christensen says your AI UX breaks because of architecture, not model quality. Dr. Jamie says your productivity breaks because of strategy, not willpower. Both are right, in their respective domains.
Durable Sessions is clearly better if your problem is AI product infrastructure. The Science-Ranked To-Do List is clearly better if your problem is personal task completion. Choosing between them is like choosing between a load balancer and a calendar app — you pick whichever matches the problem you actually have.
Which should you choose?
Choose the Christensen Durable Sessions AI UX Framework if you are an engineer, product manager, or technical leader working on an AI-powered product and your users experience dropped streams, can't resume conversations across devices, or can't interrupt a running agent. This is the right framework if your architecture audit reveals SSE-based streaming with no session persistence layer.
Choose Dr. Jamie's Science-Ranked To-Do List System if you are an individual — freelancer, manager, student, knowledge worker — who feels busy but unproductive, avoids hard tasks, has an overwhelming list, or wants a research-backed system to replace ad hoc task management.
If you happen to be an AI product engineer who also struggles with personal productivity, use both. They address different layers of your life and will never conflict.
// FREQUENTLY ASKED QUESTIONS
Can I use Durable Sessions and the Science-Ranked To-Do List System together?
Yes, but they solve completely unrelated problems. Durable Sessions fixes AI product streaming architecture. The To-Do List System fixes personal task completion. An engineer building an AI product could use Durable Sessions for their product and the To-Do List System for their own workflow. There is no integration between them — they operate in separate domains.
Which framework is better for improving my AI chatbot?
The Christensen Durable Sessions AI UX Framework is the only relevant choice. It directly addresses AI chat product issues like dropped streams, lack of multi-device continuity, and inability to cancel or steer agents mid-response. The To-Do List System has nothing to do with AI product development.
Which framework helps me stop procrastinating?
Dr. Jamie's Science-Ranked To-Do List System. It specifically targets procrastination with peer-reviewed strategies like Eat the Frog (hardest task first) and Implementation Intentions (specifying when, where, and how). It even identifies 'easy tasks first' as sophisticated procrastination — a trap that feels productive but reduces aggregate completion.
Do I need technical skills to use either framework?
Durable Sessions requires significant technical skill — you need to understand SSE, WebSockets, pub/sub architectures, and agent topologies. The Science-Ranked To-Do List System requires no technical skills at all. A pen, paper, and willingness to follow the steps is sufficient. The technical barrier is the clearest differentiator between them.
What is the single most important takeaway from each framework?
From Durable Sessions: decouple your agent layer from your client layer using a persistent session substrate so streams survive disconnections and work across devices. From the To-Do List System: attach implementation intentions (when, where, how) to every task — it is the single highest-leverage productivity move supported by 94 studies.
How long does each framework take to implement?
Durable Sessions takes days to weeks depending on your existing architecture's complexity and team size. It involves infrastructure changes, transport migration, and validation testing. The To-Do List System can be started in under an hour with a brain dump and values audit, and fully implemented within a week of daily practice.
Is the Durable Sessions framework only for companies using Ably?
No. While Mike Christensen works at Ably (a real-time infrastructure provider), the Durable Sessions concept is architecture-agnostic. The core pattern — a persistent pub/sub layer between agents and clients — can be implemented with any pub/sub system, managed WebSocket infrastructure, or custom solution. The principles apply regardless of vendor.
What tier is the best to-do list strategy according to the science?
S-tier, the highest rank, is shared by Implementation Intentions (specifying when, where, and how for every task) and Specific Time Blocking (assigning individual tasks to exact calendar slots). The lowest-ranked strategy is multitasking at F-tier — no peer-reviewed study has ever shown it improves task completion.