Lab Notes — On Sacred Cows and Engineering Reality
Every technical project develops sacred cows. Ideas that feel unquestionably correct in theory, and emotionally important to the vision of what the company should become. They persist not because they’ve survived contact with reality, but because they represent an ideal end state.
For me, one of those ideas was transfersomes.
Transfersomes are among the most effective topical delivery systems available. Properly engineered, they form highly deformable vesicles capable of transporting hydrophilic molecules and small peptides across the skin barrier with efficiency that conventional formulations cannot approach. From a scientific perspective, they are compelling. If the goal is to maximize transdermal transport without physically disrupting the skin, transfersomes sit close to the upper bound of what passive delivery can achieve.
That is why I wanted to build them.
But effective delivery systems are not defined only by their theoretical performance. They are defined by whether they can be manufactured consistently, stabilized over time, and iterated on without introducing hidden fragility into the system.
True transfersome formulations are complex. Their behavior depends sensitively on composition, hydration state, shear history, and vesicle size distribution. Achieving repeatable results often requires microfluidic mixing, tightly controlled pressure gradients, and careful post-processing. Even with that infrastructure in place, stability can be challenging, particularly during early formulation work where parameters are still being explored.
For a bootstrapped company — and especially as a solo founder — that complexity carries real cost. Microfluidic systems are not just expensive to acquire. They increase operational overhead, slow iteration cycles, and raise the risk that failures will be subtle and difficult to diagnose. Every change to composition or process can require revalidation of flow conditions, chip geometries, and stability assumptions. Pushing this too early would not be disciplined engineering.
So I made the decision to step back from transfersomes at this stage.
This was not a judgment about their effectiveness. It was a judgment about timing.
The current formulation strategy focuses on two delivery architectures that are meaningfully more effective than conventional skincare, while remaining manufacturable and stable within the constraints of an early-stage operation.
The first is a self-emulsifying nano-lipid system for lipophilic actives. This improves solubilization, surface coverage, and controlled release for oil-loving molecules without requiring complex downstream processing.
The second is a proniosome-based system for hydrophilic actives and small peptides. Proniosomes retain important advantages of vesicular delivery: protected aqueous cores, vesicle formation upon hydration, and improved transport of water-soluble molecules compared to passive diffusion. They capture much of the mechanical logic of deformable vesicles while avoiding the full burden of live transfersome manufacturing.
This is not a compromise to mediocrity. It is a deliberate choice to build a system that behaves predictably, can be tested honestly, and can support iteration.
Transfersomes make sense later, when the infrastructure to support them exists.
Making this choice also creates space for work that must happen in parallel. Right now, that includes building the chemical AI model MVP — the early-stage chemistry AI infrastructure that will eventually enable discovery of new molecules — as well as speaking directly with angel investors and beta testers to understand what the v0.1 product does well and where v0.2 needs to improve.
As a solo founder, every technical decision is also an organizational one. Attention is finite. Systems that demand constant babysitting reduce capacity to make progress elsewhere. The goal is not to build the most ambitious system possible on paper, but to build one that survives contact with reality and can evolve over time.
Transfersomes remain part of the long-term plan. When the company has the resources to invest properly — in microfluidics, process control, stability testing, and manufacturing rigor — I want to return to them and build them correctly. Not as a feature to market, but as a validated delivery platform.
For now, the work is quieter and less visible: choosing systems that scale, documenting limitations, building models that do not yet produce headlines, and having direct conversations with people willing to support slow, real progress.
This keeps the long-term goal intact without overextending the present system.
— AEON
