Building AI Capability Across a Workforce

Summary

The Problem: Organizations often adopt AI through a small number of individuals without establishing shared practices across the workforce. Teams are then left without repeatable methods for selecting use cases, choosing tools, and deciding where human judgment must remain. We addressed this problem across five professional-learning organizations. While professional learning was the setting, the same challenge arises wherever organizations need to turn individual AI use into a capability their workforce can sustain.

Our Approach: We developed a three-stage model:

  1. Building a Shared Foundation: Practical instruction on AI use cases, tools, risks, and guardrails.
  2. Embedding Practice in Existing Work: AI methods applied to deliverables participants already needed to produce.
  3. Coaching Through Implementation: Hands-on support to build, test, and transfer working tools and workflows.

The Impact: Each of the five coached organizations finished with a working artifact and the materials needed to operate or extend it. All five reported that the engagement addressed a relevant need, advanced a concrete AI goal, and provided value. Across the six embedded lessons, 3 of 4 participants rated the AI instruction “Very Useful” or “Extremely Useful,” while 9 of 10 said its depth and complexity were “Just Right”. Several workflows also became more efficient, including one in which prompt-development time dropped from eight hours to one.




The Problem

AI adoption often begins with individual experimentation; the challenge is turning that experimentation into a repeatable organizational practice.

In a 2024 survey of 1,000 executives across 59 countries, BCG found that only 26% of companies had developed the capabilities needed to move beyond proofs of concept to generate tangible value. Roughly 70% of the implementation challenges involved people and processes rather than technology or algorithms. The engagement with professional-learning organizations thus focused on making AI understandable to people (holding varying levels of technical experience), and turning that understanding into practices their organizations could sustain after the instruction ended.

Our Approach

We combined a shared foundation with practice inside existing work and coaching through implementation.


Stage 1 Shared foundation 3 practical workshops Strategy, tools, risks, and guardrails.
Stage 2 Applied practice 6 work-embedded lessons Applied examples to a work product.
Stage 3 Coached implementation 5 organizations Producing a working AI artifact.
Fig. 1: Ghamut's workforce education model combined instruction, hands-on practice, and coaching. Each stage focused more closely on the organization's workflows and produced an artifact the team could reuse.

1. Building a shared foundation across differing levels of experience: We delivered three 90-minute workshops covering AI across the professional-learning lifecycle: analysis, design, development, implementation, and evaluation. Participants practiced with the tools rather than hearing them described.

Session 1 AI for Design & Development Includes how to scale AI across the organization Tools: ChatGPT, NotebookLM
Session 2 AI for Delivery Includes when and how to adopt AI tools Tools: Coteach.ai, TeachFX, Playlab
Session 3 AI for Analysis & Evaluation Includes how to overcome barriers to AI adoption Tools: Sibme, SchedulerAI, Zapier
Fig. 2: The three foundation workshops, the questions they addressed, and the tools participants used.

The instruction was supported by market research; we reviewed 400 sector-focused AI tools, then selected approximately 20 to construct a decision guide organized by use case, setup effort, and price.


2. Embedding AI practice inside real workplace tasks: We embedded 20-minute AI lessons into an existing six-part course. Each lesson applied an AI method to a deliverable participants were already producing, and included a worked example, an activity, and direct application to the participant’s organization. The six lessons and their slides are listed in Fig. 3.

Module 1 Define product value Using AI-driven research View slides
Module 2 Identify fragile points Using AI prompting with examples View slides
Module 3 Codify a product structure Using AI self-reflection View slides
Module 4 Design a delivery workflow Using AI-assisted diagram creation View slides
Module 5 Design a pilot and manage risk Using AI ensembling View slides
Module 6 Synthesize evidence Using source-grounded AI View slides
Fig. 3: Each of the six AI lessons were designed around a real-world task participants were expected to complete.

3. Coaching teams through implementation: We provided sustained, hands-on coaching to each of five organizations in the professional learning sector. Every engagement began with a problem the organization already had; we helped define the need, built and tested a work product, explained the decisions behind it, and transferred the materials so that the team could operate and extend the result themselves. The resulting artifacts included an AI literacy framework analysis, a career-connected task generator, an AI tool for shortening and evaluating lesson plans, a curriculum-internalization chatbot, and an assessment rubric for evaluating educational technology.

The Impact

The engagement empowered five professional-learning organizations with AI tools and methods they could independently operate and extend.


All five professional-learning organizations agreed or strongly agreed that the coaching addressed a relevant need, supported adoption of new AI tools, advanced a concrete AI goal, was responsive to their context and constraints, produced useful deliverables, and provided value overall. Nearly nine in ten participants said the lessons struck the right balance of depth and complexity, while more than three in four rated them “very useful” or “extremely useful.”

775,000+ students impacted
The organizations that engaged in the program operate at national scale, supporting at least 775,000 K-12 students.

The work was recognized externally as well. One participating organization acknowledged Ghamut’s AI consultation and expertise in a research preprint, and another published an interview on applying AI to professional practice.

The AI workforce training model (shared instruction, practice within existing workflows, implementation support) we applied to professional-learning can be extended to other domains. The materials from this engagement are publicly available to help other organizations get started.

© 2026 Ghamut Corporation