From AI Pilot to Operational Learning Capability — Conversational AI in Coach Education

The Mission Problem

Moving Government Learning from “Did They Like It?” to “Can They Do It?”

Since 2018, Flatter has supported the Department of the Air Force Senior Leader Development Portfolio, serving senior leaders including O-6 officers and above, E-9 senior enlisted leaders, Senior Executive Service leaders, and joint-service participants.

Within that portfolio, the government had a specific learning-measurement challenge: move beyond attendance, completion, and participant satisfaction surveys toward evidence that participants actually learned and could demonstrate the competencies being taught. The desired direction was at least Kirkpatrick Level 2 evidence of learning.

That requirement created an opportunity for AI that was fundamentally different from using a chatbot to deliver course content. ‘Conversations that Deliver Outcomes’ became the mantra for the PersonaOPPs development team, led by Lucas Flatter.

Flatter asked whether AI could help make a human skill, professional coaching, observable, repeatable, assessable, and actionable at scale. The desire was to build better coaches that in turn would influence a Coaching Culture at the Department of War.

An Existing Program Became the Pilot Environment

In 2023, in collaboration with the DAF, Flatter selected its Coaching Culture Facilitator Course (CCFC) as an early application for PersonaOPPs.

CCFC provided an unusually strong environment for applied AI because the underlying learning architecture already existed. The course was an approximately 80-hour coach-education program serving Department of the Air Force and joint-service participants. Flatter was already an International Coaching Federation Level 2 coaching-education provider and had established curriculum, coaching faculty and program leadership, approximately 20 coaching case studies, and recognized ICF competencies and PCC markers against which coaching behaviors could be assessed. Rather than inventing synthetic AI exercises disconnected from the curriculum, Flatter converted its existing coaching case studies into interactive AI personas. The result was not an open-ended chatbot.

PersonaOPPs structured each experience around a defined persona, scenario, purpose, and result. A learner could enter a realistic coaching situation, conduct a natural-language coaching conversation with an AI persona, apply the skills being taught in CCFC, and receive structured feedback against established coaching criteria.

This approach allowed Flatter to deliver a conversational AI solution around an established mission workflow rather than introducing AI as a standalone technology looking for a use case.

The Critical Transition: Before “Coaching in the Wild”

Professional coaching creates a practical training problem. A learner can study coaching competencies, watch an instructor demonstrate them, discuss them with peers, and understand them intellectually—and still be uncomfortable conducting a real coaching conversation.

Flatter informally describes that transition as “coaching in the wild.” New coaches must eventually approach another person, ask to coach them, manage an unscripted conversation, apply the appropriate competencies in real time, and make mistakes in front of another human being.

PersonaOPPs inserted a new learning layer immediately before that transition.

Learners could practice without immediately recruiting another person. They could make mistakes privately, experiment with different questioning approaches, repeat conversations, practice outside scheduled classroom time, and build confidence before moving to human coaching.

The same need was observed with a coaching cohort associated with the Defense Intelligence Agency. Some learners were uncomfortable approaching acquaintances or neighbors to practice coaching, particularly where their professional environment made those conversations feel sensitive. AI personas provided a lower-friction first step.

The design principle was therefore not AI instead of people. It was AI before higher-consequence human practice.

How the Capability Worked

Turning Coaching Practice into an Observable Learning Cycle

PersonaOPPs was incorporated as an asynchronous practice layer alongside instructor-led education. A learner could:

1. Enter a realistic scenario — Select or be assigned an AI coaching persona derived from an established coaching case study.
2. Conduct the session — Engage the persona through a natural-language coaching conversation rather than selecting predetermined answers.
3. Demonstrate the skill — Apply coaching techniques and behaviors during the interaction.
4. Receive structured feedback — Have performance evaluated against defined coaching criteria, including ICF competencies and PCC markers.
5. Repeat and adapt — Conduct additional practice, change questioning strategies, and apply feedback.
6. Transition to human practice — Move toward coaching real people with additional practice and confidence.

This created something traditional classroom instruction alone could not provide as easily: a repeatable environment for generating observable evidence of learner behavior.

One Interaction, Three Levels of Value

Learner | Private, Repeatable Practice

PersonaOPPs gave students a place to rehearse a consequential interpersonal skill without immediately imposing on another human participant. Students could practice asynchronously outside live instruction, repeat sessions, try different approaches, receive information about how effectively they demonstrated coaching behaviors, and practice again.

The platform accumulated hundreds of hours of coaching practice, according to the program interview. Students were generally enthusiastic about having a safe practice environment, and Flatter reported no significant resistance to the AI personas.

Facilitator | Cohort-Level Learning Intelligence

The same interactions could provide information beyond the individual learner. Aggregated performance data could expose recurring competency gaps across a cohort. For example, if multiple students struggled with establishing or maintaining coaching agreements, the facilitator could identify that pattern and revisit the competency during live instruction.

That creates a closed instructional loop: Teach → Practice → Observe → Identify Gaps → Adapt Instruction → Practice Again.

Government | Evidence of Performance

For the government customer, PersonaOPPs introduced an additional category of training evidence. Traditional reporting could establish that a learner attended → completed → responded to a survey. PersonaOPPs created the ability to examine whether a learner practiced → demonstrated behaviors → received assessment → practiced again → improved.

That distinction directly addressed the portfolio’s interest in progressing toward Kirkpatrick Level 2 evidence. PersonaOPPs therefore connected three functions that are frequently separated in professional education: Practice + Assessment + Program Insight.

From Point Solution to Reusable AI Architecture

The coaching implementation demonstrates a broader productization model. PersonaOPPs does not depend on coaching because the underlying architecture is based on creating structured interactions around a persona, scenario, purpose, and desired result. In CCFC, those components were populated with coaching case studies, coaching interactions, ICF competencies, and PCC markers.

The value came from pairing the technology with a mature instructional framework and explicit assessment criteria. That distinction matters for government AI adoption.

The implementation did not require AI to replace faculty, peer interaction, human coaching, or formal credentialing. Instead, AI occupied a defined position in the existing learning architecture where automation and scale had a clear advantage: repeatable practice and structured observation. The human instructor retained responsibility for instruction and learning intervention. Human coaching remained the consequential real-world application. PersonaOPPs created the bridge between them.

Results, Scale, and the Productization Lesson

Early Evidence: From Pilot to Continued Delivery

The initial Air Force application generated several concrete indicators of adoption and utility:

• Hundreds of hours of practice. Learners accumulated hundreds of hours of coaching practice through the platform.
• Observable improvement. Flatter observed measurable improvement from approximately Day 0 to Day 90 reassessment.
• Positive learner response. Students were generally enthusiastic about having a safe environment in which to practice, with no significant resistance to the AI personas reported in the interview.
• Instructor insight. Facilitators could identify cohort-level competency gaps and use those patterns to inform live instruction.
• Government measurement value. The platform made demonstrated learner performance more visible, adding evidence beyond attendance, completion, and satisfaction.
• Transition beyond the initial pilot. The Air Force pilot was well received, and PersonaOPPs coaching practice was subsequently incorporated into later CCFC deliveries.

What Was Actually Productized?

The reusable capability was not simply an AI coaching character. Flatter combined:

• Mission context — A defined Department of the Air Force senior-leader-development requirement.
• Validated content — Existing curriculum and approximately 20 established coaching case studies.
• Human expertise — Experienced coaching faculty and program leadership.
• External standards — ICF competencies and PCC markers.
• AI interaction — Natural-language personas capable of supporting realistic practice conversations.
• Assessment — Structured evaluation of demonstrated coaching behavior.
• Iteration — The ability for learners to repeat the experience and apply feedback.
• Analytics — The ability to identify patterns extending beyond an individual interaction.

Together, those components transformed generative AI from a conversational interface into a repeatable learning capability embedded within an operational program.

A Different Model for AI in Defense Human-Capital Missions

The CCFC implementation suggests a practical path for AI adoption in missions where human judgment, communication, leadership, interviewing, counseling, facilitation, or decision-making must ultimately be exercised with real people. The high-value role for AI may not be replacing that human interaction. It may be creating a scalable environment in which personnel can rehearse it first.

For a defense organization, that model can address several persistent constraints simultaneously: access to practice partners, inconsistency in practice opportunities, learner reluctance to make mistakes publicly, limited instructor observation time, difficulty generating repeatable scenarios, and weak evidence connecting course completion to demonstrated skill.

PersonaOPPs turned practice itself into a source of data. Instead of asking only whether training occurred, the implementation creates the possibility of asking: What did the learner demonstrate? Where did the learner struggle? Did performance change after additional practice? Are multiple learners struggling with the same competency? What should the instructor reinforce?

That is the transition from AI as content generation to AI as mission infrastructure for human performance.

The Productization Pattern

1. Start with a real mission workflow. Flatter began with an established government learning program and a specific performance-measurement problem—not a generic AI demonstration.
2. Convert existing expertise into AI experiences. Approximately 20 existing coaching case studies became interactive practice personas rather than being replaced by generic AI-generated scenarios.
3. Anchor AI to an external performance standard. Interactions could be evaluated using established ICF competencies and PCC markers.
4. Put AI between instruction and consequence. Learners could practice privately and repeatedly before applying the skill with another human.
5. Turn interactions into evidence. Practice generated learner-level feedback and the potential for cohort-level performance insight.
6. Use data to improve the human system. Facilitators could identify recurring competency gaps and reinforce those areas during instructor-led learning.
7. Move from pilot to repeat use. Following a well-received Air Force pilot, PersonaOPPs coaching practice was incorporated into later CCFC deliveries.

---

Case Study Bottom Line

PersonaOPPs demonstrates a model for defense AI productization centered on practice rather than content generation.

Within Department of the Air Force coach education, Flatter took established curriculum, approximately 20 existing case studies, professional coaching standards, faculty expertise, and an identified government measurement requirement and converted them into an AI-enabled practice and assessment capability.

Learners gained a private, repeatable environment for developing human skills before “coaching in the wild.” Facilitators gained visibility into competency gaps. The government gained a pathway toward evidence of demonstrated learning beyond attendance and satisfaction.

The early implementation accumulated hundreds of hours of coaching practice, showed measurable improvement from approximately Day 0 to Day 90, was well received by the Air Force, and was incorporated into later CCFC deliveries.

The result is not a case for replacing coaches or instructors with AI. It is a case for using AI to make human performance more practiced, observable, measurable, and actionable before the real-world moment matters.

---

What Is a PersonaOPP?

PersonaOPPs are conversational AI systems built to do more than chat — they listen, think, and deliver real results. Each OPP (Operational Profile Pathway) transforms natural dialogue into structured outcomes like Smart Reports, filled forms, or training assessments. Whether for education, defense, or enterprise, PersonaOPPs make AI useful, human, and mission-ready.

The Value of Conversational AI

Most AI tools stop at answers. PersonaOPPs go further. They replace rigid forms and portals with natural dialogue, capture nuance and context that static systems miss, and deliver structured outputs that plug directly into your workflow.

How Personas Work

The “Ears” capture every detail—whether it’s a story, a requirement, or a form field—so nothing slips through the cracks. They don’t just hear words: they know when to pause, when to let you finish your thought, and how to keep the rhythm natural.

Once the Persona understands, it speaks back in ways that feel human. The “Mouth” turns raw data into dialogue—through text, chat, or lifelike voice. The result: conversations that feel less like software, more like collaboration.

The “Brain” connects the dots, reasons through problems, and ties everything into your workflows and knowledge systems. It also builds the Pathway—turning the entire conversation into a Smart Report, structured requirement, or decision log.

Always On. Always Improving.

Unlike people, Personas are available anytime to repeat scenarios, collect data, or review processes, driving measurable progress from day 1 onward. Each interaction improves skills and confidence, with metrics and Smart Reports showing both engagement and growth.

What Is an OPP? (Operational Profile Pathway)

Each OPP is a complete conversational system built for a specific purpose. The Profile defines the role—coach, analyst, investigator, or trainer. The Pathway defines the workflow—what gets captured, how it’s structured, and where it goes next.

Some OPPs are educational—focused on coaching, simulation, or guided learning. Others are operational—automating processes, intakes, or data collection. All share the same goal: turning human conversation into actionable intelligence.

Persona OPP Types

• RAG Knowledge Management — Turn static knowledge into living expertise. Personas retrieve precise facts and context on demand. Think: a 24/7 helpdesk that always knows the answer.
• Conversational Training Scenarios — Practice makes confident. These OPPs drop users into realistic role-play sessions with real-time feedback. Think: leadership coaching, interview prep, or customer-service training.
• Requirements Collection — Skip endless back-and-forth. Personas guide users through structured requirement-gathering automatically. Think: capturing technical specs for an app in a single session.
• Form-Fill Workflow — Replace rigid forms and portals with natural conversation. Think: onboarding, HR paperwork, or claims processing—done in minutes.
• RAG + Workflow Integrated — Combine knowledge retrieval and form automation for adaptive processes. Think: compliance audits, policy-driven intakes, or system configurations that need external references.
• Training & Teaching — Learn directly from expert Personas with guided lessons and feedback. Think: language learning, leadership practice, or procedural mastery.
• Interrogation / Interview — When clarity matters most. Structured questioning that extracts details efficiently and consistently. Think: security vetting, investigative interviews, or assessments.