Computation over assertion
Every result comes from a recorded run, with its inputs, parameters, code version and execution mode. A placeholder is never shown as a finding.
Advanced AI, Quantum & Technology Research
ALQERIS is the research environment of the NEO AI ecosystem. Plan research, run real computations, challenge every claim and seal each result with evidence anyone can verify.
In development - private research environment
Built into the architecture
computational labs
experiment task types
independent evidence dimensions
permission-aware LILI research tools
Figures describe the software design of ALQERIS. They are not usage or performance metrics.
Vision
AI can now produce claims faster than anyone can check them. ALQERIS is built the other way round: nothing counts until it has been computed, recorded with its provenance and exposed to challenge.
Every result comes from a recorded run, with its inputs, parameters, code version and execution mode. A placeholder is never shown as a finding.
Sources, datasets, runs and conclusions stay linked. Retracted sources are flagged wherever they are cited.
Failed hypotheses and dead ends are part of the record. They are never deleted, so no one repeats them blindly.
Review, scientific validation and signatures belong to people. The system records them, it does not imitate them.
The labs
Each lab runs computations in isolation and declares how every result was produced: local computation, simulator, fixture or live provider. Select a lab to explore it.
Measures the reliability of AI answer workflows against known ground truth, so model behaviour is scored, not assumed.
ExecutionLocal computation
Candidate algorithms compete against a fixed baseline under a protected evaluator, with mutation search and adaptive experiments.
ExecutionLocal computation
Ideal and noisy circuit simulation checked against an exact classical reference, plus resource projections for fault-tolerant scale.
ExecutionSimulator
Kernel benchmarks with warm-up, repeats, correctness gates and confidence intervals, so speed claims carry their uncertainty.
ExecutionLocal computation
A digital design flow: hardware described in code, simulated against a reference and synthesised to a generic gate library.
ExecutionSimulator
Control policies tested in physics simulation under domain shift. Simulation outcomes are labelled as such, never as physical measurements.
ExecutionSimulator
Interatomic-potential modelling, numerical solvers checked against surrogates, and symbolic regression with unit checks.
ExecutionLocal computation
Interventions and ablations on structural causal models, with estimator bias measured against a known truth.
ExecutionLocal computation
Statements checked by an SMT solver or a proof assistant. A proof is only marked verified when the checker accepts it.
ExecutionLocal computation
Holdout evaluation by a sealed evaluator. Benchmark labels never leave the vault, not even for administrators.
ExecutionSealed evaluator
A catalogue of APIs, datasets, models and tools, each with its licence, access terms and cost, ingested with checkpoints.
ExecutionFixture until activated
Quantum Lab
Quantum circuits run on ideal and noise-aware simulators and are compared with an exact classical reference. Resource projections estimate what fault-tolerant hardware would need. A result is never presented as hardware unless it ran on hardware.
Method
Every research project follows the same path. Each step leaves a trace that can be read, replayed and audited.
Research plans are written, versioned and approved before work starts. Hypotheses are stated in advance.
Each run executes in isolation, with no standing access rights, and records its inputs, outputs and execution mode.
Replications, counter-tests and sealed benchmarks try to break the result before anyone relies on it.
Results are packaged as canonical records, hashed, signed and batched into a Merkle tree.
Anyone holding a package can verify it independently, in a browser or from the command line.
Publications, IP files and partner rooms are prepared from the evidence. Nothing is ever sent or filed automatically.
Proof Ledger
Integrity is not the same as validity. ALQERIS keeps nine evidence dimensions separate and shows each one with its true status. When something has not been measured, it says so.
Merkle batch - each result becomes a leaf, every leaf is bound to one root
Illustration of the verifier output on a demonstration package.
Built on open, independently verifiable standards: canonical JSON, SHA-256, signed envelopes and Merkle trees.
Benchmark Vault
The answers to a benchmark stay inside the vault. Only the protected evaluator can read them, and it returns a score, never the labels. Not to a researcher, not to an administrator, not to an AI.
LILI, inside ALQERIS
LILI, the assistant of the whole NEO AI ecosystem, works inside every ALQERIS screen. It connects sources, experiments and results, and acts only with the rights of the person asking. Every tool call is logged.
Research tools
Illustrative conversation on demonstration data.
Security & governance
Access starts with a personal invitation link created by an administrator. There is no public sign-up.
Research leads, researchers, reviewers, auditors and partners each see and do only what their role allows.
Every action is written to a hash-chained, append-only audit journal that the application itself cannot rewrite.
Each run executes in its own sandbox with no standing role and no stored secret.
Projects are private by default. A public registry entry requires an explicit approval.
Central NEO AI administration can supervise ALQERIS, but sealed evidence and human signatures cannot be altered by anyone.
The NEO AI ecosystem
ALQERIS is one of the six NEO AI verticals, with twenty more in preparation. Each vertical keeps its own data and permissions, and all of them share the same intelligence: LILI.
+ 20 verticals in preparation
Explore NEO AIWhere ALQERIS stands
ALQERIS is in development. The research workspace, the labs and the evidence engine have been designed and built; the interactive prototype runs on demonstration data. Access will open by invitation.
Research workspace, eleven labs, evidence engine, LILI integration.
For the internal research team, with data sources activated one by one after licence review.
Partner rooms for research collaborations, by invitation.
ALQERIS in brief
ALQERIS by NEO AI is an advanced research environment for artificial intelligence, quantum and frontier technology. A research team uses it to plan work, run computations, keep every result with its provenance, challenge its own claims and seal verifiable evidence.
Not yet. ALQERIS is in development and the interactive prototype uses demonstration data. Accounts are opened by personal invitation: anyone can send an access request, and the team reviews each one.
The Quantum Lab works with simulators: ideal and noisy circuit simulation, checked against an exact classical reference, and resource projections. Any use of quantum hardware would be a separate, explicit decision, and results would be labelled accordingly.
Each result can be sealed in a proof package: a canonical record, hashed, signed and included in a Merkle batch. An independent verifier checks the package in a browser or from the command line, and shows the nine evidence dimensions separately.
LILI is the assistant of the NEO AI ecosystem. In ALQERIS it helps search sources, state hypotheses, plan experiments, compare runs, link evidence and draft reports. It acts only with the permissions of the person using it, and every action is logged.
ALQERIS is a vertical of NEO AI, developed by NEO TECH LABS TECHNOLOGY L.L.C, Dubai, a subsidiary of My NEO Group.
Access
Accounts are opened by personal invitation. Tell us who you are and what you would like to research: the ALQERIS team reviews every request and answers personally.
Already have an account?Sign in to your ALQERIS workspace.Login