Following the release of a white paper titled “The AI Mentor: Assisted Training; Customized Tutoring; Lifelong Mentor,” we caught up with APV’s Manuel Miranda, Chief AI & Growth Officer, to find out more about this unique spin on AI and the opportunity for government and industry.
Where the Idea Came From
Manuel has spent his career focused on how technology can reduce inequality — particularly in learning. Schools are chronically under-resourced, workers can’t afford private tutors, and government employees are assigned mentors who are too busy to mentor. These gaps were the seed of the AI Mentor Platform.
The idea took shape when he started working with AI four or five years ago. “Wouldn’t it be great if this could augment a human being helping someone learn?” he recalls thinking. Not a static video. Not a one-size-fits-all course. Something that could actually interact, adapt, and guide. In government, especially, the traditional mentorship model breaks down fast. The assigned mentor is the busiest person in the building, and new employees quickly learn not to ask questions that might make them look unprepared.
How the Platform Actually Works
Manuel modeled the platform on the natural arc of human learning. Phase one is the teacher phase: the AI is trained on the material and walks learners through foundational concepts, the 101 courses. Phase two is tutoring, and this is where the customization kicks in. If a learner gets 10 out of 100 questions wrong, the system identifies the underlying concepts shared by those questions and builds targeted exercises to address them. Then it retests. Everyone learns differently, and the platform adapts accordingly.
Phase three is the mentor phase, and it addresses the issue of learning to think rather than just using AI as an answer machine. The learner is now doing actual work, and the AI is not handing them answers. It is helping them develop problem-solving capability so that, at the end of the day, they can still make decisions on their own. That independence, Manuel says, matters a lot.
The Risk Nobody Is Talking About
Manuel is direct about the larger danger he sees. If people increasingly depend on AI to give them answers without understanding the reasoning behind them, the consequences go far beyond productivity. “Where does democracy go after that?” he asks.
For government, the stakes are concrete. Human-in-the-loop oversight only works if the humans actually understand what they are overseeing. Someone who rubber-stamps AI outputs 100% of the time is not a check on anything. Manuel points to national security as an extreme but clarifying example: if electronic warfare knocked out an adversary’s AI-dependent systems, and those operators had no idea how to function without them, the vulnerability would be catastrophic.
He wrote the white paper with this tension in mind. Knowing that government tends to ignore long-horizon problems, he deliberately anchored the argument in immediate, operational pain points — while still making the case that the next generation of leaders cannot afford to outsource their judgment to a machine.
A Real-World Crisis the Platform Could Help Solve
The LaGuardia incident put a human face on the cost of workforce gaps. An air traffic controller, stretched thin by chronic understaffing and managing a concurrent emergency, gave a fatal instruction. The Secretary of Transportation acknowledged that the system has seven trainees in the pipeline, trying to close the gap.
Right now, part of what those trainees learn is by shadowing. They stand next to experienced controllers and wait for situations to arise naturally. That takes years. An AI-powered training environment can simulate every scenario — rare emergencies, compounding failures, high-pressure edge cases — and compress that learning into months. More than knowledge, Manuel emphasizes, it builds confidence. When lives are in your hands, you have to know you can do the job. The AI mentor can be the backstop that gets trainees there faster.
The same logic applies across government: FDA inspectors reviewing new drugs, CMS reviewers identifying fraud. These are roles where deep expertise takes years to develop. The AI Mentor Platform puts that development on fast-forward.
Who Should Be Paying Attention
Manuel sees call center contractors as an immediate opportunity. Government call centers operate at enormous scale — one CMS contract alone runs $6 billion — and they hire in volume during enrollment periods. AI will disrupt that space faster and more thoroughly than any previous technology. The platform can train existing staff to work effectively alongside AI tools, introduce self-service capabilities so agents can focus on complex cases, and build deep expertise across a workforce rather than concentrating it in a handful of senior people.
Beyond call centers, Manuel’s advice applies to any contractor pursuing work with a training component: stop treating it as an afterthought. And for government agencies writing solicitations, he has specific guidance — require outcome-based metrics, demonstrated competency levels, and ongoing support tools. “That language will separate real solutions from checkbox training,” he says.
On Hallucinations and Getting It Right
No conversation about AI-powered training would be complete without addressing accuracy. Manuel is candid: the thing that makes AI powerful — the ability to generate a plausible-sounding answer in minutes — is also what makes it dangerous. Old technology failed loudly when it was wrong. AI fails quietly, with confidence.
The safeguard is process. Build in small pieces. Test with subject matter experts in the loop. Refine. Repeat. For APV’s firefighter training work with the Air Force and Department of War, accuracy was verified before anything went live. That rigor, Manuel says, is non-negotiable. The platform infrastructure remains consistent; what changes with each deployment is the subject matter, and that content must be validated before the system touches a learner.
APV is investing in the AI Mentor Platform as an independent capability — not built for a single client but designed for deployment across agencies and missions. Elements of the platform are already in production through a partnership supporting firefighting training across the Department of War and allied forces.
Interested in learning more? Access the white paper – short version here :Short-Version_AI Mentor_APV-White-Paper (1); full version here: 1773842292986-The AI Mentor_APV-White-Paper (1).
About Manuel Miranda, Chief AI & Growth Officer
Manuel Miranda is a visionary technology leader with over 40 years of experience driving business growth, innovation, and customer success in federal IT. As Chief AI & Growth Officer at A P Ventures (APV), he leads the company’s artificial intelligence strategy, accelerating innovation and scaling AI-powered solutions that deliver measurable impact across government missions. By showcasing the art of the possible in AI, he delivers real-world solutions that drive meaningful, measurable gains for the customer.
At the forefront of AI advancement, Manuel spearheads initiatives within APV’s Emerging Technology Lab (ETL), transforming concepts into operational capabilities. He has led the successful integration of multiple AI use cases into mission delivery and is widely recognized for his ability to align cutting-edge technologies with real-world federal challenges.
A standout achievement includes co-developing the AI-driven AR Incident Commander/AI Mentor platform with the Air Force Civil Engineer Center (AFCEC)—a pioneering training solution that earned both the 2025 ACT-IAC Innovation Champion Award.
Read the full white paper on GovWhitePapers to go deeper on the research and framework behind the platform.
