AI-native engineering

Engineered in simulation before it was built in glass.

PlastiBioFuel was conceived, modeled, and stress-tested with AI as a working partner long before the first bench experiment. More than a million simulation runs, a heat-recovery model that has been improving itself for over a year, and a founder who treats every function of the company as something a machine should help run. This page explains how.

1,000,000+
simulation runs completed before scale-up decisions
12+ months
of continuous, self-building heat-recovery modeling
1st attempt
real post-consumer bottle PET fully depolymerized in the lab
Origin of the concept

One operator, three fields of science, and a reasoning partner.

The idea did not come out of a laboratory. It came from a fuel-logistics operator who knew what a gallon of ethanol was worth, how it moved through terminals and blenders, and how much post-consumer plastic the world pays to bury. AI closed the scientific distance between those two facts.

Working through the published literature with AI as a study partner, Michael mapped three bodies of work that had never been joined: engineered enzymes that break PET down into its monomers at low temperature, immobilization techniques that keep a biocatalyst working through repeated use, and continuous-flow reactor design that keeps a solid feedstock, a fixed catalyst, and a liquid phase in productive contact. None of the three was new on its own. The invention is how they are combined, and what that combination does to the working life of the enzyme.

AI compressed the time from question to defensible answer. Literature that would have taken months to survey was synthesized in days. Competing designs were argued out numerically before anything was fabricated. Every claim was tested against what the published science actually supports, which is why the provisional patent covers the immobilization method and the reactor architecture rather than the enzyme.

  1. June 2025Concept formed: an immobilized PET-hydrolase system in a continuous-flow reactor, with the monomer stream routed to ethanol. Simulation work begins the same month.
  2. July 2025Provisional patent filed with the USPTO covering the immobilization method and reactor architecture. The heat-recovery loop starts running.
  3. November 2025PlastiBioFuel LLC formed in Texas. Federal registrations and SDVOSB verification completed.
  4. January 2026Sponsored research agreement executed with the University of North Texas.
  5. 2026Real post-consumer bottle PET depolymerized on the first attempt, then run in immobilized form with full retained activity. The simulations had the shape of the result right.
Simulation at scale

Over a million runs, from flow rate to immobilization strategy.

Before committing to a single piece of hardware, Michael built and ran simulation models that reproduce the numerical methods used in leading engineering software: computational fluid dynamics for flow and residence time, reaction-kinetic models for enzyme behavior, and process models for mass and energy balance. Then he ran them. More than one million times.

The runs attacked the process from every angle at once. Flow rates and channel geometry. Mass transfer between plastic, catalyst, and liquid. Enzyme loading, retention, and the strategies that keep a biocatalyst attached and active over many cycles. Thermal profiles. Feedstock pretreatment. Separation and recovery. Each family of runs narrowed the design space for the next, so by the time work moved to the bench, the experiments were confirming a design rather than searching for one.

That is also why the first laboratory attempt on real bottle plastic worked. The conditions had already been found in silico.

Flow

Hydrodynamics and residence time

Channel geometry, velocity, pulsing, and dead-zone elimination modeled across the operating envelope before fabrication.

Kinetics

Enzyme behavior under load

Rate, turnover, inhibition, and substrate depletion modeled against the real crystalline PET the plant will see, not idealized film.

Immobilization

Keeping the catalyst working

Loading density, retention, and reuse strategies compared head-to-head to extend working life and lower cost per ton.

Thermal

Energy balance and recovery

Temperature profiles and heat integration modeled to hold the process at low temperature with the smallest energy input.

Pretreatment

Feedstock conditioning

Size reduction, washing, and contaminant handling for mixed, dyed, and contaminated grades that recyclers reject.

Separation

Product recovery

Monomer clarification, ethanol recovery, and water recycle modeled as one system so the mass balance closes.

The heat-recovery loop

A model that has been improving itself for over a year.

Heat is the largest controllable cost in any conversion process. To get it right, Michael built a heat-recovery model that does not wait for a person to run it. It pulls from an external data source, evaluates a thermal-integration design for the process, and folds what it learned back into its own inputs for the next pass. Each cycle starts from a better place than the last.

It has been running in the background since mid-2025 with no manual restarts. What comes out is not a single answer but a continuously refined design for recovering and reusing heat across the process, and a record of how that design has changed and why. When the pilot plant is built, its thermal layout will already have a year of iteration behind it.

The data source and the model internals are confidential. The result is not: a low-temperature process designed to give back most of the energy it uses.

PullNew data from an external source
EvaluateThermal-integration design against the current process model
RecordWhat improved, what did not, and why
Feed backResults become the inputs for the next cycle
Where AI works in the company

Every function, not just the science.

A one-founder company competing with funded labs has to multiply its hours. AI is how PlastiBioFuel does that, in the lab program and in the business around it.

Research synthesis

Literature at scale

Continuous review of enzymatic depolymerization, materials science, and biofuel policy, distilled into the design decisions and citations that shape the research plan.

Process design

Reactor and flow modeling

Iteration on reactor geometry, residence time, and mass transfer in simulation, so bench prototypes start from a reasoned design rather than a guess.

Lab data

Experimental interpretation

Analysis of chromatography time-course data and turnover-frequency curves from the university laboratory, separating substrate depletion from enzyme deactivation and setting the next experiment.

Economics

Techno-economic and life-cycle models

Cost-per-gallon, credit-stack, and carbon-intensity models rebuilt as lab results land, so commercial claims track the data.

Funding

Federal and state programs

An automated sweep of SBIR, STTR, agency, and state opportunities, scored on expected value and time to first funds, feeding proposal drafting and compliance review.

Operations

Documents, contracts, and outreach

Research agreements, budgets, investor materials, and this website, produced and maintained through AI-assisted pipelines under the founder's review.

How we use it

Ground rules that keep the science honest.

  • The lab decides. Simulation proposes; chromatography, activity assays, and mass balance dispose. No claim leaves the company without a measurement behind it.
  • Novelty is scoped precisely. PlastiBioFuel did not invent PET-degrading enzymes. The work is the immobilization system, the reactor, and the integrated process, and AI-assisted prior-art review keeps that line clear.
  • A person owns every output. Proposals, contracts, and technical documents are reviewed and signed by the founder, who is accountable for them.
  • Proprietary detail stays proprietary. Materials, operating parameters, data sources, and model internals are not published here or fed to public tools. What is described on this page is the approach, not the recipe.
  • Records are kept for the record. Physical research journals are digitized and analyzed, giving the company a searchable, dated account of how each decision was made.
Questions

Frequently asked.

Does PlastiBioFuel use AI to design its process?

Yes. The process was conceived, simulated, and refined with AI as a working partner, including more than one million simulation runs before the first bench experiments. Laboratory measurement decides what is carried forward.

What does PlastiBioFuel make?

Renewable ethanol produced from post-consumer PET plastic waste, such as bottles and packaging, using an immobilized-enzyme process that runs at low temperature.

How many simulations has PlastiBioFuel run?

More than one million runs across flow, kinetics, immobilization strategy, thermal design, and separation, using numerical methods that match leading engineering simulation software.

What is the heat-recovery loop?

A self-building thermal-integration model that has run continuously for more than a year. It draws on an external data source, feeds each result back into its own inputs, and improves the heat-recovery design for the process without manual restarts.

Is the technology protected?

A provisional patent application is on file with the USPTO covering the enzyme immobilization method and reactor architecture. Specific materials and operating parameters are confidential.

Is PlastiBioFuel a veteran-owned business?

Yes. PlastiBioFuel LLC is a Service-Disabled Veteran-Owned Small Business (SDVOSB) founded and led by a 12-year U.S. Army veteran, based in Fort Worth, Texas.

How can I work with PlastiBioFuel?

Email michael@plastibiofuel.com or use the contact page. The company works with research partners, fuel buyers, waste-stream suppliers, federal program officers, and investors.

Working with us

If you build with AI, you already understand how we work.

Collaborators, program officers, fuel buyers, and investors who want to see the process up close are welcome.