Protein structure
AlphaFold predicted the three-dimensional structures of nearly all known proteins — a task that had absorbed structural biology for half a century.
AI Advanced IncResearchSpace Unknown · Eight sections
The founding thesis
Human design is bounded by cognition, not by possibility.
AI Advanced was formed on a single argument, set out in the essay Space Unknown: Designing Beyond the Human Mind. Everything the company builds follows from it.
01
Three claims.
01Design is bounded by the mind, not the universe
Human design is limited not by the set of possible solutions, but by the much smaller set a mind can conceive and evaluate in finite time with finite attention. The bounds are not personal failings; they are properties of the nervous system.
02Compute is the first instrument that reaches past it
Generative modelling, simulation and large-scale optimisation are the first tools in history capable of systematically exploring beyond that boundary — brought together for the first time in this decade.
03This is an instrument, not a productivity gain
The shift is comparable to the microscope or the telescope. Like them, the instrument invents nothing: it lets us see what was always there but never visible.
02
Nested reach.
One designer, all designers in history, and the computationally explorable — all sitting inside the space of everything that could work.
03
Five limits. Properties of the nervous system, not personal failings.
Machine search does not tire. Human attention must.
L1Working memory is small
Roughly four to seven concepts held in active consideration, against a real problem carrying twenty interacting variables. To cope, we prune — and every prune closes a door.
L2Fatigue compresses search
A designer at hour two generates more candidates than the same designer at hour eight. Decisions under fatigue tend toward the familiar — the path back to what already exists.
L3Training is a cage as much as a ladder
Internalised traditions make their forms feel like the shape of the answer rather than a shape. Mastery accelerates work inside the known region; it rarely produces movement beyond it.
L4Cross-domain integration is shallow
Excellent design lives at the intersection of mechanics, chemistry, economics, psychology and supply chain. Every designer runs a thin layer of each, and thinness limits the integrative leap.
L5Caliber expands but does not escape
A more skilled designer covers more territory, still bounded. Even the canonical figures explored a tiny fraction of what was possible; their genius was choosing well from what they could see.
The Space Unknown is not small. It dwarfs the Reachable Space in nearly every domain where it has been measured.
A protein fold no one has imagined. A circuit topology no engineer has drawn. A molecule no chemist has synthesised. A building geometry no architect has sketched.
They all exist, in principle, the way America existed in 1490 — fully real, fully unvisited.
04
The pattern is already identical across domains.
Every one of these follows the same pattern: a region of the Space Unknown was entered, and it contained something useful.
Generative structures
Spacecraft antennas, aircraft partitions and engine brackets whose forms look almost grown, outperforming their predecessors on weight, strength, or both.
Move 37
A move professionals described as alien — one no human master would have made, and yet correct.
Electrolytes & catalysts
Computational search has identified battery electrolytes and catalysts that escaped decades of human-led search.
05
The centre of gravity moves from generation to judgment.
The designer of the next generation is measured less by what she draws and more by what she chooses — and by the quality of the questions she puts to the instrument.
Framer of problems
Computational systems are extraordinary at solving problems and almost incompetent at choosing which problem to solve. Defining the goal, the constraints and the success criteria becomes the highest-leverage activity in the workflow.
Judge of candidates
When the machine proposes ten thousand viable answers, the question is which is right for these people, in this place, at this moment. That draws on values, ethics, culture and lived experience — it is irreducibly human.
Translator to the world
The optimiser does not know a building has to feel welcoming, or that a device has to fit a frightened patient's hand. Embedding tacit constraints — and noticing what the model missed — is human work, and will remain so.
06
Exploring the unknown is not unambiguously good.
None of these argue against entering the territory. They argue for entering it carefully, with humility and oversight — and with the understanding that the instrument shows us what is possible, not what is wise.
Unstated objectives
A solution can be optimal on its stated objectives and harmful on the ones we forgot to state. A drug can be effective and addictive. A network can be efficient and brittle. An algorithm can be accurate and unjust.
Verification
Intuition was a sanity check: a bridge that looks wrong probably is. When the optimiser returns something unlike any bridge ever built, that check must be replaced with rigorous simulation, testing and oversight.
Concentration
The compute and data required are not evenly distributed. Whoever controls the instrument controls the discoveries. Broad access is a political question, not only a technical one.
Meaning
Designers derive identity from creating. If creation increasingly happens through the machine, authorship has to be rebuilt around curation, judgment and intention as creative acts in their own right.
07
The thesis, in full.
Everything on this page is drawn from one essay. It is published here whole — eight sections, seven figures, thirteen pages — so the argument can be read in the author's own order rather than ours.
Essay on design and intelligence
Space Unknown
Designing Beyond the Human Mind
Bala Jeldi · June 2026
Contents
- IIntroduction
- IIThe Cognitive Boundaries of Human Design
- IIIDefining Space Unknown
- IVThe Computational Unlocking
- VDomains of Transformation
- VIThe New Role of the Human Designer
- VIIRisks and Considerations
- VIIIConclusion
On the author
Bala Anand Jeldi works on AI-driven metal additive manufacturing for aerospace, defence and other mission-critical applications, with more than two decades in additive manufacturing and magnesium alloy engineering across the space, defence and automotive industries. He holds an engineering degree from JNTU Hyderabad and is based in Carmel, Indiana.
Conclusion
The consequence is not that machines design and humans watch.
For most of history the boundary of design was the boundary of the human mind, and the greatest designers pushed it outward by a few precious metres. We are now building instruments that move it by kilometres at a time. Used carefully — pointed at the problems that matter, judged against human values, with its benefits distributed broadly — the result is that humans, for the first time, design at the scale of what is actually possible.
The point of the instrument is not to replace the designer's mind. It is to finally give the designer's mind room to breathe.