Reasoning
“Why does this decision work?”
Systems must understand dependencies, causal relationships, goals and constraints rather than relying only on statistical correlations.
Trustworthy cognitive optimization
Optimize the Impossible.
Qbitune is building a cognitive optimization engine that can reason about complex real-world goals and constraints, abstract them into formal models, generate adaptive plans, and find better decisions across classical, hybrid and quantum computing.
ReasonUnderstand goals, constraints and changing context
PlanBuild adaptive strategies across multiple time horizons
VerifyReturn decisions that can be checked and explained
QBITUNE / COGNITIVE ENGINE
VERIFIED
Illustration of the pipeline Qbitune is building. Not a product screenshot — the system is at research stage.
DEEP REASONING◆
FORMAL ABSTRACTION◆
ADAPTIVE PLANNING◆
VERIFIABLE DECISIONS◆
HYBRID COMPUTE◆
EUROPEAN DEEP TECH◆
DEEP REASONING◆
FORMAL ABSTRACTION◆
ADAPTIVE PLANNING◆
VERIFIABLE DECISIONS◆
HYBRID COMPUTE◆
EUROPEAN DEEP TECH◆
01 / THE COGNITIVE GAP
AI can generate answers.
Real operations need valid plans.
Modern AI is exceptionally good at recognising patterns and generating content. But complex real-world decisions demand something more: understanding constraints, reasoning about consequences, planning over time and adapting when reality changes.
Abstraction
“What is the actual problem?”
Real environments contain thousands of details. Intelligent systems must identify the concepts, relationships and constraints that actually matter.
Planning
“What should happen next?”
Useful intelligence must generate long-term strategies, anticipate contingencies and revise plans as circumstances change.
02 / THE QBITUNE ENGINE
From messy reality
to verified decisions.
-
01
Understand
Interpret the operational context, objectives and available evidence.
-
02
Reason
Identify dependencies, conflicts, uncertainty and the consequences of alternative decisions.
-
03
Abstract
Construct a compact world model containing goals, entities, resources, constraints and relationships.
-
04
Plan
Generate hierarchical strategies, contingencies and candidate plans across multiple time horizons.
-
05
Optimize
Translate suitable subproblems into formal representations and route them to the appropriate optimization engine.
-
06
Verify
Check hard constraints, evaluate uncertainty and produce a traceable explanation of the resulting decision.
-
07
Re-plan
Monitor new information and adapt the plan when the environment changes.
The environment changes, and the loop runs again — re-planning feeds straight back into understanding.
03 / COGNITIVE CAPABILITIES
Intelligence beyond
pattern matching.
The three capabilities overlap: a plan is only useful if the abstraction is right, and an abstraction is only right if the reasoning behind it holds.
Deep reasoning
Combine learned representations with explicit knowledge and formal reasoning to understand dependencies, constraints and possible consequences.
Deep abstraction
Transform heterogeneous real-world information into reusable concepts and formal world models that capture what actually matters for a decision.
Deep planning
Build adaptive hierarchical plans that remain useful when environments, resources and objectives change.
04 / TRUSTWORTHINESS
A good answer is not enough.
It must be a decision you can trust.
For high-value industrial decisions, optimization cannot be a black box. Qbitune is being designed to combine learned intelligence with explicit constraints, formal verification and auditable decision traces.
Formal constraints
Hard operational rules remain explicit and testable.
Verification
Candidate plans can be automatically checked before execution.
Explainability
Show which objectives, constraints and events influenced a decision.
Uncertainty
Represent uncertainty explicitly and evaluate robust alternatives.
Provenance
Maintain a trace of the data, model state and reasoning process behind each plan.
Designed for research into transparent, accountable and human-centred cognitive AI.
05 / INDUSTRIAL TESTBED
One factory.
Millions of possible plans.
A world that never stops changing.
A manufacturing site has to coordinate all of this at once — and every one of these moves depends on the others.
OrdersMachinesOperatorsSkills
MaterialsMaintenanceEnergy pricesDeadlines
Urgent order added
Machine failure
Energy price spike
- Understand the event
- Update the world model
- Determine affected goals and constraints
- Generate alternatives
- Optimize the new plan
- Verify feasibility
- Explain what changed
How we intend to evaluate it
Qbitune has no demonstrated performance results to report yet. These are the measures the research is designed around.
Plan feasibilityObjective qualityRobustness
Re-planning latencyCompute resourcesExplanation fidelity
Problems of this shape
Manufacturing is the primary demonstrator, but the difficulty is the same wherever goals, constraints and resources all move at once. These are the kinds of problem Qbitune is designed for — described, not claimed as delivered work.
01
MANUFACTURING
Production scheduling
Find better sequences for jobs, machines and production lines while respecting capacity, maintenance, materials and delivery deadlines.
Optimize for Throughput · Lead time · Machine utilization · Energy
Keeps changing Machine failures · Urgent orders · Material delays
02
ENERGY
Grid & energy optimization
Coordinate generation, storage, distributed resources and demand under cost, reliability and renewable-energy constraints.
Optimize for Cost · Stability · Renewable utilization · Flexibility
Keeps changing Price spikes · Weather · Demand peaks
03
LOGISTICS
Routing & fleet optimization
Continuously determine better routes and resource assignments across vehicles, warehouses and delivery networks.
Optimize for Distance · Fuel · Time · Fleet utilization
Keeps changing Traffic · Cancellations · Vehicle breakdowns
04
OPERATIONS
Workforce & resource scheduling
Match people, skills, equipment and tasks across complex operational constraints.
Optimize for Coverage · Cost · Productivity · Service levels
Keeps changing Absences · Skill availability · Demand shifts
05
AEROSPACE
Cargo & operations optimization
Optimize loading, allocation and scheduling decisions while respecting physical, safety and operational constraints.
Optimize for Payload · Fuel · Revenue · Operational efficiency
Keeps changing Load changes · Weather · Slot reassignments
Further transfer domains under consideration:
Supply chainsMobilityScientific discovery
06 / COMPUTE ORCHESTRATION
Use the best engine
for the problem.
Qbitune is not based on the assumption that quantum computing is already superior for every optimization task. The cognitive layer should remain independent from the machine used to search for solutions.
Quantum when it helps. Classical when it wins.
Cognitive model
Formal problem
Qbitune solver fabric
ClassicalMathematical programmingConstraint programmingMetaheuristics
HybridDecompositionQuantum-inspiredHeterogeneous workflows
QuantumAnnealingGate-based algorithms
Candidate plans
Verification
Best valid decision
07 / RESEARCH FOUNDATION
From quantum optimization research
to cognitive optimization.
Qbitune co-founder Ernesto Acosta's research demonstrates one concrete example of the underlying capability: taking a difficult optimization problem, reformulating it into a QUBO representation and solving it through a hybrid quantum workflow with recursive refinement.
Qbitune expands that research direction into a broader scientific question: can intelligent systems automatically understand complex real-world decision problems, build formal representations, generate adaptive plans and select the right computational tools to solve them?
RESEARCH EXAMPLE
QUBO-based training for VQAs on quantum annealers
Ernesto Acosta · Guillermo Botella · Carlos Cano
Quantum Machine Intelligence, Springer (2026)
doi:10.1007/s42484-026-00441-1
An early research example — not the limit of Qbitune's technology.
Read on SpringerLink ↗08 / THE RESEARCH QUESTION
Can machines learn to formulate the problem before solving it?
Optimization engines are powerful when humans provide clean variables, objectives and constraints. Generative AI can interpret messy information but cannot reliably guarantee valid long-term plans.
Qbitune is investigating the missing layer between them.
- Human expert
- Manually formalizes problem
- Solver
- Static solution
- Real-world context
- Cognitive reasoning
- Automatic abstraction
- Formal model
- Adaptive planning
- Heterogeneous optimization
- Verification
- Continual re-planning
From solving formal problems to understanding which formal problem should be solved.
09 / RESEARCH ROADMAP
From scientific hypothesis
to integrated cognitive system.
This is a research plan, not a status report. Qbitune claims no current technology readiness level.
- 01
Foundations
Formal representations and optimization research.
- 02
Cognitive prototype
Reasoning, abstraction and planning architecture.
- 03
Trustworthy integration
Verification, explainability and uncertainty.
- 04
Real-world validation
Industrial planning demonstrator.
- Target
Integrated cognitive AI prototype
Validated at TRL4. This is the research target — not current status.
10 / EVALUATION
If it cannot be measured,
it cannot be trusted.
A proposed evaluation framework for the system we are building.
Reasoning
Logical and constraint validity.
Abstraction
Transfer and model correctness.
Planning
Plan feasibility and objective quality.
Adaptation
Performance after unexpected events.
Trust
Explanation fidelity and robustness.
Compute
Time, energy and computational resources.
We aim to contribute reusable benchmarks, evaluation protocols and interoperable interfaces for trustworthy cognitive planning.
11 / COLLABORATION
Built to work with
a European research ecosystem.
Interoperability
Open interfaces for cognitive models, planners and solvers.
Benchmarks
Shared tasks and reproducible evaluation.
Joint pilots
Composable technology that can participate in larger multi-system demonstrations.
FAIR research
Research outputs designed around Findable, Accessible, Interoperable and Reusable principles where applicable.
Potential synergies include European AI research infrastructures, testing facilities, HPC and quantum initiatives.
12 / EUROPE
Europe should own the intelligence
that runs on its infrastructure.
Europe is investing in AI, high-performance computing and quantum technologies. Qbitune is being built in Europe to help create the cognitive software layer connecting complex real-world decisions with this emerging computational infrastructure.
European deep tech
Trustworthy AI
Hybrid compute
Open interoperability
Industrial sovereignty
TEAM
Built by scientists
and entrepreneurs.
Carlos Rodríguez
Co-Founder & CEO
Company building, product strategy, commercialization, partnerships and financing.
Ernesto Acosta
Co-Founder & CTO · Quantum & Optimization Research
Optimization modelling, quantum algorithms, QUBO formulations, hybrid quantum computing and research architecture.
QUBO-based training for VQAs on quantum annealers ↗Building the multidisciplinary team
Join Qbitune ↗WORK WITH US
Bring us a planning problem that machines still struggle to solve.
We are looking for research collaborators and industrial partners working on complex decision environments with changing goals, constraints and resources.