--- title: "Stage 3 — Apply AI: Prove the Value" description: "Apply generative AI to real work, measure time savings, quality and costs, and use human oversight to decide whether further investment is justified." url: "https://www.capbera.com/stage3-apply-ai" published: "2026-10-27T17:38:19Z" authors: ["Lorelei Cheung"] categories: ["Strategy & Performance"] tags: ["AI Readiness","AI Maturity Model"] --- # Stage 3 — Apply AI: Prove the Value Transition from experimenting with Generative AI to implementing it to improve a task. At this stage, you will: choose a task to augment, define success measures, estimate costs, test solutions, and evaluate results. ![photo of blocks stacking to show how AI helps build a strong foundation](https://www.capbera.com/assets/uploads/d2ac2996ad0d5db24b2db8655de3b562.png) # Your Result At ***Stage 3: Apply AI*,** your organization begins using chat-based, Generative AI to complete real work tasks identified during [Stage 2: Explore AI](https://capbera.com/stage2-explore-ai). Instead of experimenting with AI out of curiosity, you're now testing whether it can improve the way work gets done. **The best tasks to start with are typically:** - Low-risk and non-critical - Time-consuming and repetitive - Completed primarily by one person - Easy to evaluate for quality, accuracy, and time savings **Examples of tasks well suited to this stage include:** - Drafting emails - Generating images and other visual assets - Creating reusable templates to standardize work products - Transcribing meetings and compiling summaries - Documenting processes and procedures - Planning work and clarifying priorities - Brainstorming solutions to organizational challenges - Preparing work products such as presentations and reports - Preparing industry briefings and research summaries ### Build Skills Through Practice As your organization begins applying Generative AI to everyday work, employees develop their skills in prompting, evaluating outputs, and directing AI toward useful results. Training on Generative AI tools will help employees get more consistent results and improve the speed and quality of their work. Chat-based, Generative AI such as ChatGPT, Claude, Gemini, and Perplexity are increasingly used to augment employees' work. Multiple employees—and potentially entire departments—may use these tools regularly. When approaching a task, employees begin asking, "Could AI help me do this better?" The goal isn't simply to use AI more often. It's to determine where AI can make a meaningful difference to the work your organization needs to accomplish. ### Measure the Value As Generative AI becomes part of everyday work, your organization may begin seeing meaningful time savings. Employees can redirect some of that time toward higher-value activities, organizational priorities, and work that requires human judgment. To understand whether Generative AI is delivering real value, begin documenting: - **Time saved:** How long did the task take before AI, and how long does it take now, including the time needed to review and correct the output? - **Quality:** Does the work meet your organization's standards for accuracy, completeness, and usefulness? - **Cost:** Do the benefits justify subscription fees, training, implementation, and the time spent using and reviewing AI? - **Consistency:** Can employees achieve reliable results across repeated attempts? These measures help your organization decide whether to continue, refine, or discontinue involving AI in a particular task. Not every organization needs to progress to more advanced AI. For some, Generative AI tools may provide sufficient value to meet their needs. The right level of investment depends on the work to be done, the results achieved, and the organization's priorities. ### Manage Risk Through Human Oversight At this stage, employees generally interact with AI through chat-based, Generative AI tools and a person reviews the outputs before they are used. This keeps implementation relatively straightforward and provides an opportunity to develop sound review practices before introducing more complex systems. However, human review does not eliminate risk. Employees still need clear guidance on handling confidential information, verifying factual claims, checking outputs for errors, and following applicable security and privacy requirements. The objective is to build confidence in using AI responsibly while developing the skills and practices needed for broader adoption. # What to Avoid **Start small and evaluate before expanding.** Choose one or two worthwhile tasks rather than trying to apply AI across the organization all at once. Taking on too many tasks makes it difficult to measure results, identify problems, and determine where AI is genuinely adding value. Avoid adopting AI simply because a tool is available or other organizations are using it. Every task where AI is used to support work should address a real need, have a clear measure of success, and justify the time and money invested. **Increase investment only when the benefits justify the cost.** A slow, deliberate approach to AI adoption helps your organization look beyond the hype and focus on real results. By evaluating what works before investing further, you can direct resources toward AI tools that genuinely improve how work gets done and avoid spending money on technology that delivers little value. # Focus Areas | **Activity** | **Why It Matters** | | --- | --- | | **Select one to two specific, worthwhile task(s)** | Gives implementation a clear purpose and a defined problem to solve. | | **Apply AI to real work using chat-based tool** | Moves AI from experimentation into practical work without introducing unnecessary technical complexity. | | **Develop skills in prompting and directing AI** | Guides employees on how to direct AI effectively and assess whether its outputs are accurate and useful. | | **Provide training and guidance** | Builds employee confidence and establishes consistent expectations for responsible use. | | **Begin using paid AI tools where justified** | Expands available capabilities when free tools are insufficient and the expected benefits justify the cost. | | **Define success measures before implementation** | Establishes how the organization will evaluate improvements in time, quality, and cost. | | **Compare results against existing costs & performance** | Determines whether AI delivers meaningful improvements over the current way of working. | | **Establish repeatable output-review practices** | Identifies errors, protects sensitive information, and maintains quality standards. | | **Decide whether to continue, refine, or discontinue applying AI** | Ensures that future investment is guided by evidence rather than enthusiasm or sunk costs. | | **Identify opportunities for more specialized AI tools** | Determine whether specific departmental needs warrant exploring dedicated tools or more advanced AI capabilities, such as finance and sales. | # Progressing to Stage 4: Integrate AI **The goal of Stage 3 is to prove the value of AI in practice.** By testing real tasks, developing employee skills, measuring results, and establishing responsible review practices, your organization builds the evidence and experience needed to decide what should become part of its standard way of working. The milestone is a shift from experimentation to purposeful, repeatable use. AI is producing useful work, employees know how to use it effectively, and the organization has evidence that its investment is delivering value. ### Does Every Organization Need to Progress to Stage 4? **No. And that's perfectly fine.** The right level of AI adoption depends on your organization's needs, capacity, values, and priorities. For some organizations, Generative AI will provide all the support they need. Investing in more advanced AI simply because it's available may introduce unnecessary costs and risks. Your organization may decide to remain at Stage 3 for several reasons: - **Security and risk considerations:** Your systems, security requirements, or internal policies may make it impractical to give AI agents access to your organization's systems and data. You may decide that the additional risks and oversight requirements aren't worth the potential benefits. - **A preference for direct human involvement:** Your organizational culture or philosophy may prioritize human judgment and involvement in every work product or process. You may want AI to support employees without allowing it to act autonomously or complete entire processes independently. - **Sufficient organizational capacity:** Generative AI may already be helping employees complete their work efficiently. If your organization has the capacity it needs, there may be little reason to invest in more advanced tools. - **Limited need for additional capacity:** If your organization isn't growing, doesn't anticipate a significant increase in workload, or has no pressing operational constraints, the additional expense of more advanced AI may not be justified. ## Is Your Organization Ready for Stage 4: Integrate AI? For other organizations, Stage 3 will reveal opportunities that Generative AI alone cannot fully address. If your organization is struggling with capacity, managing increasing workloads, or looking for ways to improve entire processes rather than individual tasks, it may be time to explore more advanced AI capabilities. Before moving to **Stage 4: Integrate AI**, your organization should be able to: - **Use Generative AI to support recurring work:** Employees regularly use AI to produce work products, solve problems, and complete routine tasks. - **Demonstrate measurable value:** Your organization can identify meaningful benefits, such as time saved, improved quality, or increased employee capacity. - **Justify its investment:** Paid tools and training are supported by demonstrated results and actual organizational needs. - **Apply AI policies in practice:** Your organization has developed, discussed, and tested its policies for responsible AI use in real-world situations. Employees understand how to apply them, and lessons learned have informed improvements. - **Verify the quality of AI-generated work:** Outputs have been reviewed for accuracy, completeness, and adherence to organizational standards. Errors have been identified, lessons have been learned, and review procedures have been refined to improve reliability. - **Identify opportunities for more advanced AI:** Your organization has considered whether specialized tools or AI agents could help address specific departmental needs, such as those in finance or sales. - **Build a knowledge library to train AI Agents:** Like human employees, AI agents need organizational context to perform effectively. You will need to give AI Agents access to a knowledge library that includes your organization's: background, processes, policies, systems, and other relevant information so they can carry out their responsibilities in alignment with how your organization operates. These foundations help your organization make informed decisions about where to invest next and how to manage the additional complexity that may come with more advanced AI. # What's Next? At **Stage 4: Integrate AI**, your organization begins moving beyond using AI to assist with individual tasks and starts exploring how it can become part of established workflows and processes. You'll begin to: - **Use AI agents to support entire processes or projects**, where appropriate, rather than relying solely on AI to assist with individual tasks. - **Integrate Generative AI and AI agents into departmental workflows**, based on the needs and readiness of each department. - **Define clear responsibilities** for human employees, Generative AI, and AI agents, including where human review and approval are required. - **Evaluate the costs, benefits, and risks** of integrating AI into more complex work, ensuring that further investment remains justified. The focus shifts from proving that AI can help to determining how it can reliably support the way your organization operates. **Ready to take the next step?** Learn more about [Stage 4: Integrate AI](https://capbera.com/stage4-integrate-ai).