Durable Sessions vs Agentic Law Framework: Which to Use?
// TL;DR
These two frameworks solve entirely different problems and should never be evaluated as alternatives. If you are building or fixing the real-time streaming and connectivity layer of an AI-powered product, use the Christensen Durable Sessions AI UX Framework. If you are planning or scaling AI transformation for a legal team or law firm, use the Legora Agentic Law Transformation Framework. There is zero overlap in their domains — one is infrastructure architecture, the other is legal operations strategy.
// HOW DO THEY COMPARE?
| Dimension | Christensen Durable Sessions AI UX Framework | Legora Agentic Law Transformation Framework |
|---|---|---|
| Best for | Engineering teams building AI chat or agent-driven products with streaming UX | Legal teams, law firms, and in-house legal departments deploying or scaling AI |
| Core problem solved | Fragile AI streaming connections that break on disconnect, lack multi-device continuity, and prevent live user control | Moving legal teams beyond first-generation AI efficiency tools toward autonomous, agentic execution of legal work |
| Domain | AI product infrastructure and real-time architecture (domain-agnostic) | Legal technology and legal operations (deeply domain-specific) |
| Complexity | Moderate — requires understanding of streaming protocols (SSE, WebSockets), pub/sub, and session management | High — requires understanding of legal workflows, multi-jurisdictional operations, change management, and a six-layer platform architecture |
| Time to apply | Days to weeks for an architecture audit and redesign; implementation varies by stack | Months to years across three transformation phases; sustained organisational change management |
| Prerequisites | An existing AI product with a streaming architecture (SSE, WebSocket, or polling) to audit | A legal team with baseline AI tools deployed and strategic ambition to move beyond efficiency gains |
| Output type | Architecture redesign: gap map, session layer design, transport upgrade plan, and validation tests | Transformation roadmap: maturity diagnosis, AOS layer map, Skills architecture, service model redesign, and change management plan |
| Creator background | Mike Christensen (Ably) — real-time infrastructure and streaming architecture expert | Max Junestrand (Legora) — legal tech CEO building an agentic operating system for law |
| Key abstraction | Durable Sessions — a persistent, shared pub/sub layer decoupling agents from clients | Agentic Operating System (AOS) — a six-layer vertical stack from foundation models to enterprise governance |
| Multi-agent relevance | Directly addresses multi-agent streaming: sub-agents write to shared sessions, eliminating orchestrator relay bottlenecks | Uses sub-agents for parallel legal workstreams (e.g., due diligence) but focuses on task orchestration, not transport |
What does the Christensen Durable Sessions AI UX Framework do?
The Christensen Durable Sessions AI UX Framework diagnoses and fixes broken AI chat and agent-driven product experiences caused by fragile streaming architectures. It identifies the Single-Connection Trap — the default pattern where an AI response stream is coupled to a single client connection — and provides an architectural redesign centred on Durable Sessions.
A Durable Session is a persistent, shared pub/sub channel that sits between the agent layer and the client layer. Agents write events (token chunks, tool results, status updates) to the session; clients subscribe to it. This decoupling unlocks three foundational capabilities: Resilient Delivery (streams survive disconnections), Continuity Across Surfaces (sessions follow users across tabs and devices), and Live Control (clients can steer, interrupt, or cancel agents mid-generation).
The framework is particularly strong at exposing the SSE Resume-Cancel Conflict — the irresolvable ambiguity in SSE-based architectures where closing a connection could mean either "I disconnected, buffer and let me resume" or "I pressed stop, cancel generation." It prescribes replacing SSE with bidirectional transport when live control is needed and provides concrete validation tests for the redesigned architecture.
What does the Legora Agentic Law Transformation Framework do?
The Legora Agentic Law Transformation Framework is a strategic and operational blueprint for legal teams moving beyond first-generation "Legal AI" (faster-same-work) into Agentic Law — a fundamentally different model where AI agents autonomously plan, execute, review, and complete legal work with human oversight integrated throughout.
It provides a three-phase progression model: Phase 1 (Leverage) handles high-volume, low-complexity work; Phase 2 (Differentiation) restructures service delivery so a team of five competes like a team of 50; Phase 3 (Reinvention) productises legal knowledge into scalable, AI-native workflows and new business models.
The framework maps any legal team's current technology against the six layers of the Agentic Operating System (AOS): Foundation Models, Agentic Harness, Data and Integrations, Context and Knowledge, Legal Skills and Capabilities, and Enterprise Security and Governance. It introduces domain-specific constructs like Skills (plain-English instructions encoding a team's expertise), Lists (structured, executable matter management replacing Excel and Word), Monitors (real-time regulatory horizon scanning), and the Legal Engineer role — former lawyers turned AI experts who drive adoption and change management.
How do they compare?
These frameworks operate in entirely different domains and solve fundamentally different problems. The Durable Sessions framework is infrastructure-level — it concerns itself with how AI-generated events travel from agents to clients reliably, across devices, and with user control. It is domain-agnostic and applies to any AI product with a streaming UX: coding assistants, customer support bots, research tools, or any agent-driven interface.
The Legora framework is domain-specific and strategic — it concerns itself with how legal organisations restructure their operating models, workflows, and value propositions around agentic AI. It assumes the infrastructure works and focuses on what legal work gets done, how it is organised, and how the organisation transforms.
Where they share surface-level vocabulary — both mention "agents," "orchestrators," and "sub-agents" — the meaning differs. Durable Sessions addresses how sub-agent outputs are transported to clients without bottlenecks. Legora addresses how sub-agents are tasked across legal workstreams like corporate, IP, and employment due diligence.
The Durable Sessions framework is faster to apply (days to weeks for an architecture audit), while the Legora framework describes a multi-year organisational transformation. Durable Sessions requires engineering expertise in streaming protocols and real-time infrastructure. Legora requires deep legal domain expertise, change management capability, and strategic planning.
Which should you choose?
Choose the Christensen Durable Sessions AI UX Framework if you are an engineering or product team building an AI-powered product and your streaming UX breaks under real-world conditions — disconnections kill responses, users cannot switch devices mid-conversation, or your stop button does not work reliably. This framework gives you a concrete architectural pattern to fix these problems permanently.
Choose the Legora Agentic Law Transformation Framework if you are a legal team leader, law firm strategist, or legal operations professional asking "what comes after we have automated the basics?" It provides a structured path from incremental efficiency to full reinvention as an AI-native legal organisation.
If you are building a legal AI product and need both a robust streaming architecture and a domain-specific operating model, you might apply both — Durable Sessions for the transport layer, Legora's AOS for the legal workflow layer. But they answer different questions and should be evaluated independently, not as competitors.
There is no scenario where one substitutes for the other. A legal team using Legora still needs reliable streaming infrastructure (which Durable Sessions addresses). An engineering team using Durable Sessions gains nothing from Legora unless they are specifically building for the legal vertical.
// FREQUENTLY ASKED QUESTIONS
Can I use the Durable Sessions framework for a legal AI product?
Yes. Durable Sessions is domain-agnostic and applies to any AI product with streaming UX, including legal AI products. It solves the transport and connectivity layer. You would pair it with a domain-specific framework like Legora for legal workflow design, but the streaming architecture patterns apply universally regardless of vertical.
Is the Legora framework only for law firms or does it work for in-house legal teams too?
It works for both. The framework explicitly addresses in-house legal departments, law firms, and legal services businesses. The three-phase transformation model and AOS architecture apply regardless of team type, though specific use cases and economic models differ between in-house teams and firms.
Do these two frameworks compete with each other?
No. They solve entirely different problems in different domains. Durable Sessions is an infrastructure architecture pattern for AI product streaming. Legora is a legal operations transformation framework. They could be used together — Durable Sessions for the real-time transport layer of a legal AI product, Legora for the legal workflow and operating model layer.
What technical skills do I need to apply the Durable Sessions framework?
You need familiarity with streaming protocols (SSE, WebSockets), pub/sub architectures, and client-server connection management. The framework is aimed at engineering and product teams who build AI-powered applications. No domain-specific legal, medical, or other vertical expertise is required.
How long does it take to implement the Legora Agentic Law Transformation Framework?
The full transformation spans months to years across three phases. Phase 1 (Leverage) can begin delivering value within weeks as high-volume tasks are automated. Phase 2 (Differentiation) and Phase 3 (Reinvention) require sustained change management, Legal Engineer capability, and iterative organisational redesign over a longer horizon.
What is the biggest pitfall of the Durable Sessions framework?
Building resume and replay logic inside the agent itself rather than in a dedicated session layer. This couples agent code to connection management, scales poorly, and defeats the core principle of agent-client decoupling. The session layer — not the agent — should handle all reconnection, buffering, and event replay.
What is a Legal Engineer and why does the Legora framework require one?
A Legal Engineer is a former lawyer who is also a deep AI expert. They understand legal workflows, risk tolerance, and organisational culture. The Legora framework considers them essential because AI transformation in legal requires change management alongside technology — deploying, tailoring, and partnering with legal teams to bring the agentic system to life.
Can I use SSE instead of WebSockets with the Durable Sessions framework?
Only if you do not need live control (stop buttons, steering messages, mid-generation follow-ups). SSE is one-way, so closing a connection is ambiguous — it could mean disconnect or cancel. If live control is required, the framework prescribes replacing SSE with a bidirectional transport like WebSockets to resolve this fundamental conflict.