About

AECE Technologies LLC is an Ohio-based, founder-led research and development company combining field-informed product development with AI-assisted research, engineering, testing, and design.

Firsthand operating knowledge translated into practical product development

John David Luptak II

Founder, AECE Technologies

Ohio-based · Founder-led · Formed in 2025 · AI-native R&D

Why I Founded AECE Technologies

I founded AECE Technologies after seeing two problems collide: an essential physical industry was still coordinating critical operations through paper records and human memory, while artificial intelligence powerful enough to improve those operations could not yet be trusted with consequential decisions.

For more than two decades, I worked across the aggregate and material-hauling industry and its connected on-site construction field operations. Throughout that industry, I saw dispatch orders, haul routes, load tickets, loading cycles, payroll records, and billing information move through clipboards, handwritten logs, phone calls, radio instructions, and disconnected systems.

Those inefficiencies affected far more than paperwork. Poor coordination could send trucks in the wrong direction, leave loaders waiting, slow quarry production, create unnecessary equipment movement, produce incorrect records, and contribute to costly mistakes. Dispatch efficiency and quarry safety were closely connected because drivers, equipment operators, field personnel, and office staff were all acting on the same operating information.

My first attempt to address those problems was Stone Direct, an artificial-intelligence concept for dispatch, haul routing, and quarry and mine-site safety.

Testing that concept exposed a much larger problem. The AI could provide confident directions to addresses that did not exist. It could generate convincing but incorrect information or stop when it was uncertain. In an ordinary conversation, a wrong answer might only be frustrating. In a physical operating environment involving trucks, heavy equipment, people, schedules, and money, it could be expensive or dangerous.

That experience changed the question I was trying to answer.

It was no longer enough to ask how AI could improve dispatch. I began asking how any business could trust an AI whose information, decisions, or actions might affect the real world.

I used generative AI to bridge technical knowledge gaps and accelerate my research. I developed a structured process involving multiple AI systems rather than accepting the conclusions of a single model. I would formulate a problem and a possible solution, use one system to test it, another to try to break it, and others to challenge assumptions and compare the results. I used those systems as research instruments—not as final authorities—and made the architectural and strategic judgments myself.

That process led to a different idea:

Instead of trying to create one supposedly perfect AI, why not create infrastructure that different AI systems could operate inside—allowing AI to use its capabilities while preventing an unsafe, unsupported, or unauthorized action from reaching the outside world?

That question led to PRISM, AECE’s first-generation zero-trust AI-enforcement infrastructure.

PRISM was not an AI model. It was infrastructure surrounding the AI. It was designed to govern how AI logic entered a controlled environment, how proposed outputs and actions were evaluated, what evidence was recorded, and whether an action was permitted to affect an external system.

The underlying objective was straightforward: allow AI to use its capabilities inside a controlled environment while releasing only actions that satisfied defined authorization, integrity, and safety conditions. Unsupported or unsafe actions were intended to remain contained.

I separately developed ARC as another research architecture. PRISM and ARC were intentionally designed to remain separate rather than becoming one combined system.

We ran simulators to evaluate the architectures. The simulations showed that the underlying controls could function, but they also revealed a serious practical limitation. PRISM’s comprehensive, continuously active structure involved too many interacting enforcement layers to operate efficiently in its original form. It would have introduced excessive latency and would have been too expensive to deploy as a practical commercial system.

That did not invalidate the concept. It identified the next engineering problem.

I began redesigning the work around smaller, purpose-specific mechanisms that could enforce defined controls when required without keeping the entire original infrastructure continuously active. That evolution led AECE into broader research involving dependable AI infrastructure, machine accountability, controlled outputs and actions, evidence-based authorization, ephemeral enforcement, multi-domain governance, memory architecture, and accountable machine operations.

AECE Technologies grew around that research. The company is developing technical specifications, simulation-tested architectures, software prototypes, commercial products, and a growing intellectual-property portfolio. Multiple U.S. patent applications have been filed, with non-provisional and additional filings in preparation.

DAISY is the first major real-world product family through which AECE is beginning to apply selected parts of this work.

DAISY LT is being developed as an AI-assisted field-to-office product for load tracking, invoicing, payroll preparation, reporting, and business administration. DAISY DI is a separate product focused on dispatch intelligence and operational coordination. Together, they return AECE’s research to the operating problems that originally inspired the company.

But the underlying AECE technology is not limited to DAISY, material hauling, or one AI model. The broader objective is to develop infrastructure that can potentially support different AI models, agents, and automated systems wherever consequential outputs and actions must remain connected to current information, authorized control, and accountable records.

I founded AECE Technologies to turn firsthand operational insight into advanced AI products for businesses that are often overlooked by the technology industry—and to continue developing the deeper infrastructure needed for AI to earn trust in the real world.

FROM ONE PROBLEM TO A BROADER QUESTION

Useful operational AI required more than an intelligent model.

My original goal was to build AI for dispatch, operational coordination, and safety in demanding material-handling environments. Pursuing that goal showed me that useful operational AI depends on more than an intelligent model; it also depends on trustworthy context, evidence, defined authority, and controls over real-world actions.

That realization expanded AECE’s work beyond dispatch and into the broader conditions that make AI-assisted systems useful and accountable in consequential environments.

Context · Evidence · Authority · Action control

AI-ASSISTED DEVELOPMENT

Generative AI became the force multiplier.

Generative AI became the force multiplier that made this work possible. I use multiple AI systems across research, software development, testing, analysis, and product design.

AI accelerates the work; company direction, operating judgment, business decisions, and final responsibility remain human-led.

AECE created and is developing the Daisy product family from this combination of field experience and AI-assisted development.

DAISY LT, under active development and field testing, records and manages the field-to-office work. DAISY DI, in development, is the separate dispatch-only product that coordinates where the work goes next. The products can operate together using shared operating information, but they are separate products.

DAISY LT and DAISY DI combine practical material-hauling workflows with AECE-developed infrastructure designed to keep important AI-assisted outputs and actions connected to current operating information, authorized review, and accountable records.

Both products begin with AI assistance and authorized human review; their longer-term direction is bounded autonomy under company-defined permissions, rules, thresholds, current-state checks, and exception controls.

AI — Accelerates the work

Human — Leads direction and final responsibility

Two connected goals.

Today, AECE connects two goals: building practical products for overlooked businesses and researching infrastructure for more dependable AI-assisted systems.

01 Practical products for overlooked businesses

02 Infrastructure for more dependable AI-assisted systems

AECE’s work begins with real operational problems and extends beyond any one industry or AI model. The goal is to make advanced operational intelligence more useful, accountable, and accessible.