Something important is happening inside companies, and it is moving faster than most learning programs.

EY’s 2026 AI Pulse Survey recently found that 87% of senior leaders at organizations investing in artificial intelligence (AI) say they have fully deployed or are piloting programs to develop in-house, AI-built software for internal use. Yet 72% say their organizations are facing challenges with that software that are slowing progress.

The ability to build is spreading faster than the skills required to build well. AI is making it possible for employees in marketing, sales, operations, finance and human resources to create agents, workflows and automations that perform real work and have real access. The technical barrier between having an idea and building something useful from it has dropped dramatically.

For learning leaders, that creates a new mandate. AI literacy is no longer enough. Organizations need AI builder readiness — the capabilities employees need to create systems that perform work, interact with other systems and access data on their behalf. Let’s examine five capabilities that should be at the center of that effort.

1. AI Opportunity Thinking

Traditional problem-solving starts with a problem and searching for a better answer. AI builders need to go further. They need to recognize where AI changes what is possible in the first place.

The opportunity is often bigger than making an existing task faster. AI can change the shape of the work itself: manual processes can become continuous, multiple handoffs can become coordinated workflows, and work that depends on someone remembering to do it can happen automatically.

For example, an employee might use AI to draft each customer response faster. An AI builder asks a different question: Could an AI system monitor incoming requests, determine what kind of response is needed, retrieve the right information and draft or route the response automatically?

For learning and development (L&D), the goal should not be to train employees to sprinkle AI into existing tasks. It should be to help them recognize where AI can fundamentally change how work gets done, and decide what is worth building.

2. AI Building Fluency

Recognizing an opportunity is one thing. Turning it into something real requires a different level of fluency.

Employees increasingly need to understand the building blocks of modern AI systems, including models, agents, instructions, tools, memory, workflows and automation. More importantly, they need to understand how those pieces work together to perform a specific job.

This is not traditional technical training. Employees do not need to understand every layer of the technology or know how to code. They need enough fluency to make intelligent choices about what to build, how to structure it and when they have moved beyond their own expertise.

Knowing how to use AI effectively is valuable. Knowing how to assemble AI capabilities into something that performs useful work is a different skill, and one that is quickly becoming important beyond engineering.

3. Context Engineering

Context is one of the most important emerging skills in AI yet is often the least understood. Context engineering is the practice of giving an AI system the information, instructions and resources it needs to perform a task effectively. That might include business objectives, organizational priorities, source material, examples, historical decisions, permissions, available tools, policies and information created by other agents.

As AI systems become more capable, deciding what they should know becomes part of the design. AI builders need to think deliberately about what an agent can access, what it should remember, what information should follow a piece of work and what context should remain private or restricted.

This is a new frontier, but not an alien one. People already know that a colleague performs better with the right background, expectations and resources. AI builders need to learn how to make that context explicit, persistent and usable.

4. Governed Judgment

As AI systems become more autonomous, builders are making decisions that once sat largely with engineering, security and IT. What data can an agent access? Which tools can it use? What actions can it take independently? Where is human approval required?

For employees, governance means understanding and applying the boundaries that determine what an AI system is allowed to see, do and decide. That could mean limiting an agent to approved data sources, requiring human review before it sends an external communication, preventing it from changing records in a business system or knowing when a request needs to be escalated.

Deloitte’s 2026 State of AI research found that only 1 in 5 organizations has a mature governance model for autonomous AI agents. That gap becomes more consequential as agents move from generating information to taking actions across business systems.

Every builder does not need to become a security expert. But they should understand the blast radius of what they build: what the system can access, what it can change, who could be affected and what happens when something goes wrong. Governance, in other words, has to become a building skill, with permissions, approvals and boundaries built in from the start.

5. Operational Thinking

AI builders need to understand what happens after something starts running. What did the agent do? Why did it do it? Where did it fail? Is it still producing useful results? What happens when the surrounding systems or processes change?

Operational thinking means treating an AI system as something that needs to be monitored and maintained, not a one-time project. Builders need to test what they create, observe how it behaves in real situations, evaluate the quality of its outputs and know when it should be improved, restricted or stopped.

As AI adoption expands, organizations may have thousands of agents, workflows and automations operating across teams and systems. Building creates an ongoing responsibility for what was built.

Assess Readiness Across the Builder Lifecycle

This also calls for a different kind of skills assessment. Instead of asking whether employees understand AI or have completed training, L&D leaders can assess readiness across three stages of the builder lifecycle:

• Discover: Can employees identify a meaningful opportunity for AI, define the problem and decide whether building something is the right solution?
• Build: Can they combine instructions, context, tools and organizational guardrails to create something that performs useful work reliably?
• Operate: Can they monitor what their systems are doing, evaluate the results, identify problems and improve or stop them when necessary?

Now map the five capabilities against those stages. The gaps you see become much more useful for design. For instance, a marketing organization might be excellent at discovering opportunities but weak in context engineering. An operations team might have strong building fluency but inconsistent governance. Another group may be creating useful agents but have little visibility into what happens once those agents begin operating.

That tells a learning leader something an AI literacy score cannot: where employees are getting stuck as they move from AI users to AI builders. That is a much better starting point for curriculum design.

Make Building a Learning Practice

AI has dramatically lowered the technical barrier to creating software and automation. It has not eliminated the skills required to do it well. Training programs need to evolve accordingly.

AI education should become hands-on, role-specific and progressive, with employees learning by building things connected to their actual work — from recognizing opportunities to building useful systems and operating them responsibly.

The measures of success should evolve, too. Course completion can tell L&D who attended. Tool usage can show adoption. Neither proves that someone is ready to build.

That requires more than another AI course. Building itself needs to become a learning practice, with employees developing their skills by creating, testing, operating and improving real AI systems in the context of their work. When almost anyone in an organization can become an AI builder, knowing how to build well becomes an organizational capability.