Certified or labelled — never guessed

Solve the problems
that don't fit.

Describe a hard or large decision problem as JSON — a graph to cut, a tour to shorten, an assignment to place. HexStellar returns the answer with a receipt you can verify yourself.

Free trial compute units. No card.

HXS · Discovery engine

A search space collapses into one answer.

Cortex

The exact mind for the age of generative AI.

A language model imagines the next word — a superpower for writing, and exactly the wrong instrument the moment the answer has to be right: the shortest route, the one stable configuration, the portfolio that satisfies every rule at once. There, a confident guess is worse than none, because it looks like truth. Cortex is the half of the brain that does not guess. Pair them and you get the architecture the moment was waiting for.

The model

Imagines the question

Brings language, context and creativity — it understands what the human wants and formulates the problem. Fluent, flexible, open-ended.

Cortex

Answers it

Returns an answer that is optimized, reproducible, and labelled for exactly how far to trust it — certified when it earned the word, a hypothesis to verify when it did not. Never a guess dressed as truth.

The exact core is the half that compounds: every problem shape it learns to speak is permanent, and work already proven is reused instantly instead of recomputed. Progress here does not evaporate — it accumulates. That is why a few months of this can be worth years. And we do not yet know everything it will be used for: the moment your problem can be phrased as “pick the best combination under constraints,” it already has a door here. The most interesting user may be an AI itself — give an agent a partner that solves the exact part reliably, and it stops hand-waving and starts building.

Built for the combinatorial ones

Problems where the search space grows like 2ⁿ or n! and brute force is off the table.

Optimize

QUBO / Ising ground state — the core hard-optimization primitive.

MaxCut

Partition a weighted graph to maximize the cut.

TSP

The shortest closed tour over a distance matrix.

QAP

Assign facilities to locations, minimizing flow × distance.

Select

Pick exactly K of M, maximizing value minus redundancy.

Spectrum

Eigenvalues, ground energy and spectral gap of a matrix.

You don't get one use. You get every shape.

The hardest problems in the world reduce to the same few mathematical shapes — a graph to cut, a tour to shorten, an assignment to place, a set of constraints to satisfy. Formulate your problem in that shape and HexStellar returns a fast, optimized answer. What was an exclusive club of exascale compute is now one API call any AI agent can make. A few doors below — there are thousands more.

Logistics & mobility

Route fleets, pack trucks, sequence stops

Last-mile delivery, ride dispatch, warehouse slotting, container loading — rebalanced live, not planned the night before.

TSP · VRP

Drug discovery

Pick the molecules worth the wet lab

Fragment and feature selection over billion-scale libraries, binding-site combinatorics — billions collapsed to the few worth testing.

QUBO · selection

Advanced materials

Search lattices for the structure you want

Stable atomic arrangements, alloy ordering, crystal-defect placement — the configuration search at the heart of materials design.

Ising · QUBO

Finance & risk

Build portfolios, net trades, track indices

Cardinality-constrained selection, index tracking, trade netting — fast enough for a desk that reprices in real time.

QUBO · MIQP

Physical AI & robotics

Place, schedule and coordinate on-device

Task allocation across arms and agents, motion sequencing, cell layout — light enough in RAM to run on the robot itself.

QAP · assignment

Energy grids

Commit units, balance loads, site storage

Unit commitment, demand-response scheduling, EV-charging orchestration — the mixed-integer heart of a grid deciding now.

MILP · QUBO

Synthetic biology

Design pathways and gene circuits

Codon and pathway optimization, enzyme assignment, minimal-media design — state the trade-off; the engine returns the optimum.

QUBO · MILP

Quantum-class physics

Find the ground state classically

Spin-glass ground states, MAX-2-SAT, planted-optimum recovery — the exact kernel quantum hardware is built to chase.

Ising · QUBO

Chip & hardware design

Place, partition and route silicon

Cell placement, netlist partitioning, layer assignment — the layout choices that decide how fast and cool a chip runs.

QAP · MaxCut

Telecom & networks

Assign spectrum, place caches, route flow

Frequency assignment, cache placement, network partitioning — keep a live network optimal as demand moves across it.

MaxCut · coloring

Operations & scheduling

Roster shifts, sequence jobs, slot rooms

Crew and nurse rostering, job-shop sequencing, exam timetabling — the constraints most teams still solve overnight.

MILP · QUBO

Your idea

Anything that reduces to a hard choice

If it is “pick the best combination under constraints,” it already has a shape the engine speaks. Bring one.

you decide

Every one is a graph, a matrix or a set of constraints in disguise. Give the engine the shape — it returns the answer.

The impossible is only impossible until someone with a little creativity tries it.
Brayon Pieske · Founder, HexStellar
Rules mode

Write the business rules. Nothing else.

Modelling a hard problem used to mean hand-coding the mathematical model for weeks. Now you send plain JSON — "type": "capacity_limit", "k": 5 — and the engine enforces every rule as a hard constraint and returns a valid answer, in milliseconds. Nine rules take the engine from raw optimization to the language of whole industries.

choose_one

exactly one of a set is picked

choose_exactly

pick exactly K of a set

capacity_limit

at most K of a set — the weight cap

mutual_exclusion

these two can never be picked together

requires

if this is picked, that must be too

identical

these must share the same choice

different

these must differ

force_true

this one is fixed on

force_false

this one is fixed off

Billion-dollar verticals it models instantly

Molecular docking

Virtual screening for new drugs — rotate a ligand through millions of 3D fits against a protein pocket.

choose_one · mutual_exclusion

Last-mile & fleets

Assign 5,000 packages across hundreds of trucks, minimizing distance without breaking any truck’s load.

choose_one · capacity_limit

Portfolios & hedge funds

Pick exactly 15 of 5,000 stocks, blocking two names from the same risk sector, maximizing return under hard risk rules.

choose_exactly · different

Rostering & aviation

Build doctor, nurse or pilot shifts under labour law — pairings that must hold, nights that block mornings, fixed days off.

requires · mutual_exclusion · force_true

5G antenna allocation

Colour a whole city’s radio channels so no two nearby antennas share a frequency and drown each other out.

mutual_exclusion

Cloud & data centers

Pack VMs and containers onto servers — a rack holds at most 4 heavy VMs; DB and API must sit together for latency.

capacity_limit · identical

You send the rules; the engine returns an answer that satisfies every one of them — zero violations, in milliseconds. No modelling by hand. The limit is the imagination of your AI and your customers.

Every answer says what it is

Below the exact cap an answer is a certified optimum, proven by exhausting the space. Above it, it is a hypothesis — strong, and labelled as such. We never claim an optimality we did not earn.

certified optimum heuristic — verify
response
{
  "answer": [0, 1, 0, 1],
  "cut_value": 4,
  "certainty": "certified optimum (proven by exhaustion)",
  "receipt": "a91f2c3b8d04"
}