Every role has a layer AI can do and a layer it can't. The substitutable layer framework is how the AI Career Index splits them.
Every professional role can be decomposed into two layers: the substitutable layer and the durable layer. The substitutable layer is the portion of the work AI tooling can already perform at acceptable quality and near-zero marginal cost. The durable layer is the portion that AI cannot perform because it requires judgment, accountability, physical presence, or trust that the role structurally routes through a human. The AI Career Index methodology uses this split as the central framework for measuring exposure.
How the split is defined
The framework categorises every task in a role into three buckets. Routine production work sits in the substitutable layer: drafting, coding, analysis production, standard communication, templated reporting. Strategic judgment work and relational trust work sit in the durable layer: exception handling, accountability decisions, client-facing relationship, licensure-backed opinion, live crisis response.
The typical knowledge-worker role in the index has 40 to 60 percent of task time in routine production, 20 to 40 percent in strategic judgment, and 10 to 30 percent in relational trust work. Roles above 60 percent routine are in the high-exposure band. Roles below 25 percent routine are in the low-exposure band.
Why this split matters more than job titles
Two people with the same job title can have very different substitutable layer shares. A marketing manager who spends 70 percent of the week on campaign production has a very different reading from one who spends 70 percent on brand strategy and stakeholder alignment. The substitutable layer framework makes this distinction legible where the job title hides it.
This is also why every role page in the directory publishes specific substitutable tasks and durable skills. The lists are not decoration. They are the operating definition of where AI is reaching for the role and where it is not.
The framework as a migration tool
The practical use of the framework is career migration planning. If your substitutable layer share is above 60 percent, the migration path is to move task time from routine production into strategic judgment or relational trust. Every role page in the directory publishes a specific six-month plan for making this shift, tailored to the role's band and category. See the executor-to-architect migration playbook for the underlying pattern.
The goal is not to eliminate the substitutable layer — a small routine production share is normal in most roles. The goal is to keep the substitutable layer below the threshold where AI substitution is economically competitive with continued employment. Empirically, that threshold is around 50 percent for most knowledge-worker categories.
How the substitutable layer changes over time
The substitutable layer is not static. It expands as AI tooling matures. Five years ago, drafting a regulatory filing required deep domain expertise. Now, first-pass filings are drafted by AI and reviewed by a human. Work that was in the durable layer is now in the substitutable layer. The pace of expansion varies by function: text-heavy work moves fastest; physical work moves slowest; regulated work moves slower than unregulated work because licensure constrains adoption.
The implication: a role that scored 30% substitutable in 2023 may score 50% substitutable in 2025 because the same work is now within AI reach. The structural protection that worked two years ago is not the structural protection that works now.
How to read your substitutable layer
Audit your last 10 hours of work. Categorise each task: routine production, strategic judgement, or relational trust. Calculate the substitutable percentage. The audit takes 30 minutes and produces a baseline reading that you can re-run every 6 months to track the migration.
Most professionals are surprised at the result on the first audit. Routine production usually exceeds the gut estimate by 10-20 percentage points. Once visible, it becomes actionable: the highest-leverage move is to compress the routine layer with AI tooling and redeploy the reclaimed hours upward.
A caveat on the model
The framework is a structural model, not a forecast. It measures what AI can substitute given current capability, not what AI will substitute given firm-level adoption decisions and labour-market response. The substitutable layer score is a planning figure, and the migration plan is an actionable response. Neither is a prediction of inevitable replacement.
Related: the compression curve, authority bands, Compression Exposure Index, executor-to-architect migration.
Run your structural assessment
The AI Career Index is free to take. Get your structural snapshot in 10 minutes, including your durability score, archetype, and projection.
Take the Free Assessment