Applied Research Institute

Applied AI for the physical world.

Independent research into the AI-safety problems that matter — open, verifiable, and neglected by the labs racing past them.

Latest — ColludeBench pilot: 20 rounds × 3 conditions, RFC 3161-timestamped & pre-registered

Independent California 501(c)(3) · EIN 41-4991887 · research published open-source

2
Research frontiers
8+
Open-source RL environments
RFC 3161
Timestamped pre-registration
100%
Re-derivable results
Flagship research

Measuring multi-agent collusion — before it measures us.

As AI systems begin to negotiate, price, and coordinate on our behalf, they can learn to collude in ways no single-agent test would catch. ColludeBench is a pre-registered, timestamped benchmark that measures it directly.

ColludeBench

A pre-registered, RFC 3161-timestamped benchmark for multi-agent LLM collusion. Agents run repeated pricing games across controlled conditions; the harness measures whether a network of models compresses prices supra-competitively and whether communication amplifies the effect.

Pre-registered RFC 3161 timestamp Open source Reproducible
~35%
Price compression measured in the pilot pricing game
FAF ≤ 1.0
Communication suppressed — did not amplify — collusion in the tested condition
Model-dependent
Effect size varied by model family — the open question the benchmark exists to map

Pilot / Stage-2b results. Full protocol and data in the repository.

Open research

Research that anyone can re-run.

Beyond the flagship benchmark, HHA maintains a suite of open-source reinforcement-learning environments for high-stakes clinical and embodied decision-making — every one published, versioned, and installable.

Clinical RL

AnestheSim

RL for anesthesia dosing.

Clinical RL

VentiSim

RL for mechanical ventilation.

Clinical RL

GlucoSim

RL for glucose management.

Clinical RL

OncoSim

RL for radiation-therapy planning.

Clinical RL

CardioSim

RL for cardiac electrophysiology.

Neural interface

NeuroSim

Brain–computer interface environments.

Embodied AI

VascularSim

Microbot vascular navigation.

Molecular

PeptideGym

Peptide-design environments.

Programs

Fundable research tracks.

Each program is a TRL-mapped track with defined milestones — structured so a funder can back a specific, verifiable outcome rather than a vague mandate.

01

Evaluation & adversarial testing

Reproducible evaluation frameworks with binary-testable criteria, red-team methodology, and multi-perspective stress-testing.

02

Multi-agent safety & governance

Failure-cascade propagation, unintended coordination, and constraint erosion — with policy translation for democratic oversight. ColludeBench is the anchor program here.

03

Embodied AI systems

AI-driven control across scale tiers, and the manufacturing engineering that bridges laboratory prototypes to deployable systems.

04

RL for medical decision-making

Open reinforcement-learning environments for clinical applications, distributed as installable packages.

05

Methodology infrastructure

Cross-domain pipelines from research question to re-derivable artifact, with reference designs per category.

Funding status: grant applications in progress with the Long-Term Future Fund and the Cooperative AI Foundation. Grants fund the neglected open research the commercial side will not.
The Institute

Independent by design.

HHA is a small, independent research group. The people are the method — and the governance is deliberately built so the research answers to the evidence, not to a commercial incentive.

Haedar Hadi
Haedar Hadi
Lead Principal Investigator
MS Computer Science (Boston University). Leads benchmark development, multi-agent system architecture, and reproducible evaluation pipelines.
Hass Dhia
Hass Dhia
Co-Principal Investigator
MS Biomedical Sciences. AI-infrastructure architect. Leads experimental design, adversarial evaluation methodology, and analysis, with domain expertise in physiology and pharmacokinetics.
Ahmed Dhia
Ahmed Dhia
Director of Manufacturing
Design for manufacturability, production scaling, and quality systems (ISO 13485). Leads dataset curation and the path from reference design to buildable hardware.

Independence & governance. HHA is an independent California 501(c)(3). Research is published open-source, and every primary result is designed to be re-derived by a third party in a different toolchain.

Where research produces a commercializable reference design, that design is licensed at arm's length to commercial partners, and any such license is subject to conflict-of-interest review. Commercial revenue never sets the research agenda.

AI for physical systems

Reference designs.

The second frontier: putting learned intelligence onto the smallest possible hardware. These are research artifacts — published methods, not products.

Current reference design
On-device motion AI

Cassette

A gait-coaching reference design engineered against an 8 KB on-sensor (ISPU) memory budget, pairing on-sensor classification with a phone-side coaching model. On-device fit is currently predicted from a datasheet-derived model; hardware bench validation is pending.

How commercialization works

Reference designs are licensed at arm's length to commercial partners under conflict-of-interest review. Royalties flow back to support open research — but the commercial path only ever picks up designs the research has already published. HHA sells nothing here.

Licensing inquiries

Methodology

Every claim resolves to an artifact.

The through-line across both frontiers: nothing is asserted that a third party cannot re-derive. That discipline is the product.

01

Criteria-driven

Research questions decompose into discrete, binary-testable criteria before any investigation begins.

02

Pre-registered

Hypotheses, endpoints, and stopping rules are locked and RFC 3161-timestamped before data collection.

03

Cross-toolchain

An independent implementation in a different language reproduces every primary numeric result.

04

Open by default

Papers, code, protocols, and timestamps ship publicly — the record is the receipt.

Support the work

Fund research the labs are racing past.

HHA is grant-funded and independent. The open, verifiable safety work here exists because someone chose to fund the neglected question instead of the crowded one. That can be you.

Also open to: research collaboration · reference-design licensing · joining the team