Two dimensions of AI readiness are systematically underassessed: process and talent. Organizations invest heavily in technology and strategy while underestimating how much process fragmentation and talent scarcity will constrain their ability to execute. The result is programs that are well-funded and well-intentioned but chronically underperforming.


Process Readiness

AI Surfaces What Organizations Did Not Know They Had

The most uncomfortable property of AI deployment is diagnostic. When you apply AI to a business process, you see the process as it actually operates, not as it was designed to operate. What emerges is frequently disturbing.

Processes that leadership believed were standardized turn out to be executed in dozens of different ways across geographies, business units, and teams. Decisions that appeared rule-based turn out to depend on institutional knowledge held by a handful of individuals. Workflows that seemed efficient turn out to have redundant steps, approval loops with no clear owner, and exception handling that exists nowhere in any system.

AI does not fix these problems. It exposes them. And then it amplifies them. An AI system trained on fragmented process data learns fragmented process behavior. A workflow automation built on a poorly documented process automates inconsistency at scale.

The uncomfortable truth

Process audits are almost never included in AI readiness assessments. Organizations evaluate their data, their technology, and their talent. They rarely ask: "Is this process documented, standardized, and stable enough for AI to learn from and operate within?" The omission is expensive.

The Fragmentation Pattern

This pattern appears in nearly every enterprise AI engagement at scale:

  1. Leadership identifies a high-value process for AI deployment (invoice processing, customer onboarding, demand forecasting)
  2. The AI team begins discovery and finds the process is executed materially differently across regions, business units, or teams
  3. The variation is not documented anywhere. It exists as tribal knowledge
  4. The AI project cannot proceed without first standardizing the process
  5. Process standardization requires cross-functional alignment that takes months
  6. The AI project timeline extends. Stakeholder confidence drops.

The same process executed dozens of ways across geographies is not a data problem or a technology problem. It is a process governance problem that predates AI and that AI has now made impossible to ignore.

Where this shows up most

High-fragmentation processes include: procurement approval workflows, customer contract negotiation, employee onboarding, regulatory reporting, and any process that was "standardized" by a policy document but never enforced at the system level.

Process Readiness Assessment

For each candidate AI use case, assess the underlying process across these dimensions:

DimensionQuestions to Ask
DocumentationIs the process formally documented? Is documentation current and accurate? Do practitioners recognize it as reflecting how they actually work?
StandardizationIs the process executed consistently across all instances, or does significant variation exist? Is variation intentional (for legitimate reasons) or accidental?
MeasurabilityAre process inputs, outputs, and cycle times measured? Is baseline performance data available?
StabilityHas the process been stable for at least 12 months? Is it likely to change significantly in the next 12 months?
OwnershipIs there a named process owner with authority to make and enforce standardization decisions?

Scoring:

ScoreStateImplication
1-2 across most dimensionsProcess is not readyProcess remediation must precede AI deployment. Estimate 3-9 months of process work.
3 across most dimensionsProcess is partially readyPilot in the most standardized instance first. Use pilot to drive standardization in others.
4-5 across most dimensionsProcess is readyAI deployment can proceed. Monitor for process drift post-deployment.

What AI-Ready Processes Look Like

A process is AI-ready when:


Talent Readiness

The Scale of the Gap

Only 20% of companies rate their talent as highly prepared for broad AI adoption, down two points on the previous year and the lowest score of any readiness dimension Deloitte measures (Deloitte, 2026). This is not a pipeline problem that will resolve itself in 18 months. It is a structural gap that requires deliberate intervention.

The conventional response is to hire data scientists. This is necessary but insufficient. The talent gap in enterprise AI is broader, and the missing roles are less visible.

The Talent Gap Is Not What You Think

Organizations tend to underinvest in the roles that determine whether AI programs succeed at scale, while overinvesting in the roles that determine whether individual models perform well.

Roles that are well-understood (and usually prioritized):

Roles that are systematically neglected:

AI process architects. These are practitioners who understand both business process design and AI capability. They translate between what AI can do and what business processes need. They conduct process audits, identify automation opportunities, and design the human-AI handoff points that determine whether a system is usable. Most organizations have none.

AI program leads. Deploying AI at scale requires program management discipline that most IT or data science leaders do not have. AI program leads manage cross-functional dependencies, stakeholder alignment, change management integration, and the portfolio-level view of initiative sequencing. The role sits at the intersection of transformation program management and AI domain knowledge.

AI governance specialists. Risk, compliance, and audit functions need practitioners who understand AI-specific risks: model drift, data provenance failures, algorithmic bias, and regulatory exposure. Most governance functions are staffed with generalists who are learning AI on the job.

Business translators. The capability gap between technical AI teams and business unit leaders is real and costly. Business translators bridge it. They are not data scientists and they are not business analysts in the traditional sense. They understand AI well enough to identify valuable use cases and communicate technical constraints to business stakeholders, and they understand the business well enough to identify where AI creates genuine leverage.

The Emerging AI Generalist

PwC calls it the "rise of the generalist": a move toward broader, outcome-focused roles, creating demand for people who understand a wide enough range of tasks to oversee agents and align their work with business goals (PwC, 2026). PwC expects the shape of the workforce to change with it, toward an hourglass in knowledge work, with strong junior and senior tiers and fewer middle roles, and toward a diamond in frontline work, where more mid-level people are needed to orchestrate agents.

Read that as a talent specification, not a forecast. The generalist role it describes is a domain expert who can identify, scope, evaluate and supervise AI within their own function without routing every decision through a central AI team. That is what a working AI literacy program produces, and it is where the leverage is: generalists reduce the bottleneck on the central team and accelerate adoption inside the business units.

The Access vs. Usage Gap

A critical distinction that most talent readiness assessments miss: access to AI tools is not the same as effective use of AI tools.

Workforce access to sanctioned AI tools grew by 50% in a single year, from under 40% to just under 60% of workers. Among those with access, fewer than 60% use AI in their daily workflow, a pattern essentially unchanged on the previous year (Deloitte, 2026). Access widened. Usage did not. The gap between the two is where AI adoption programs fail.

The reasons for the gap are consistent across organizations:

Closing the access-usage gap

The organizations making fastest progress on the access-usage gap are doing three things: embedding AI tools in specific workflows rather than offering them as general-purpose tools, having managers use AI publicly and share outputs in team settings, and measuring usage as a management metric alongside business outcomes.

Talent Readiness Assessment

Role CategoryQuestions
Technical (data science, ML engineering, data engineering)Do we have enough capacity to run 3-5 concurrent production AI projects? Is there a career path that retains senior practitioners?
Program and processDo we have AI program leads with transformation experience? Do we have process architects who can conduct process audits for AI readiness?
Governance and riskDoes legal, compliance, and risk have practitioners who understand AI-specific risks? Is there an AI governance function with teeth?
Business translationDo business units have AI translators who can identify and scope use cases without depending entirely on the central team?
AI generalistsAre we actively building AI fluency in domain roles? Do we measure usage, not just access?

Interpretation:

Missing program + governance + translationAI generalists emergingTechnical onlyPrograms succeed technically, fail organizationallyTechnical + program + governance + translationScalable AI adoption

A talent portfolio that covers only technical roles will produce AI systems that work as engineered and fail as organizational interventions. The full talent portfolio is required for transformation.

Building the Talent Capability

Short-term (0-6 months):

Medium-term (6-18 months):

Long-term (18+ months):


The Integrated View

Process and talent readiness are connected. Organizations with strong process documentation and standardization can onboard AI generalists into business functions because those generalists have stable processes to work with. Organizations with deep AI talent can accelerate process standardization because they have the practitioners to conduct rigorous process audits.

The failure mode is treating both as separate workstreams that can be addressed sequentially. They are not. The organizations that scale AI successfully address process and talent in parallel, as integrated components of the same transformation program.



Sources

  1. Deloitte AI Institute. "State of AI in the Enterprise: The Untapped Edge." January 2026. Survey of 3,235 leaders across 24 countries. Figure 12 (preparedness by dimension) and the worker access and pilot expansion findings.
  2. PwC. "2026 AI Business Predictions." 2026. Prediction on the rise of the generalist and agent orchestration.

For the complete source list and methodology, see Sources & Methodology.