AI systems
Agents, tooling & LLM workflows
Coding agents, model routing, workflow DSLs, and LLM-oriented language experiments for real developer environments—including public work such as BeanAlis and BeanAlis GUI.
Explore AI workHong Kong software company
Japality Limited builds inspectable scientific tools, AI agents, and store apps. Selected systems are public on GitHub. Computational project work is paid, scoped, and written down before it starts.
Hong Kong · App Store & Google Play · info@japality.com
Rust · custom-panel classifier
./microbiolens analyze \
--r1 reads_R1.fastq.gz \
--r2 reads_R2.fastq.gz \
--db my_lib/db.json \
--index my_lib/cascade.idx \
--method cascade
Custom-panel classification of partner-provided reads.
Rust · in silico specificity
japality-primer matrix \
--forward ATCGATCGATCGATCG \
--reverse GCTAGCTAGCTAGCTA \
--max-mismatch 3 \
L_plantarum:ref_lp.fasta \
L_buchneri:ref_lb.fasta
In silico specificity check of candidate primers.
TypeScript · terminal agent
beanalis
# multi-step repo tasks
# MCP tools · hybrid routing
# verify gates before emit
Bring your own OpenAI-compatible API. Model output still needs review.
What we do
We ship products, publish selected code, and take on paid software projects with research and industry partners. Each engagement starts with a clear problem and a practical deliverable.
AI systems
Coding agents, model routing, workflow DSLs, and LLM-oriented language experiments for real developer environments—including public work such as BeanAlis and BeanAlis GUI.
Explore AI workScientific software
Sequence and NGS utilities such as MicroBioLens and japality-primer, plus chemistry and visualisation apps. Computational analysis of partner-provided sequence files.
How a sequence project is scopedProduct engineering
A published portfolio across science, education, utilities, and games on the App Store and Google Play—plus selected open-source repositories on GitHub.
Browse appsScientific software
A paid software engagement using tools we already ship. The full worked example is on the research collaboration page.
Input
After a written agreement, the partner shares authorised FASTQ or other sequence files for computational analysis.
Software
Custom-panel classification of reads against taxa the partner chooses. Public Rust CLI, configured for the project.
Software
In silico specificity review of candidate primers—PCR simulation and multi-genome matrices.
You get software configuration, a computational report, and method notes another engineer can rerun. Fees and deliverables are set in writing.
Read the worked exampleSystems already public
A subset of our work is published under open-source licences. Most mobile apps remain proprietary store products.
Rust · Metagenomics
Custom-panel microbial shotgun / 16S classifier with cascade indexing and convergence early-stop.
Rust · Primer analysis
In silico primer specificity checker—IUPAC-aware PCR simulation, multi-genome matrices, and FASTQ scanning.
TypeScript · AI agent
AI coding agent for the terminal—multi-step tasks, MCP tools, hybrid routing, and verify gates.
Mobile apps
Consumer and productivity apps are distributed through official app stores. Store listings govern availability, pricing, and product-specific privacy.
Science
On-device RNA-seq alignment and splice-site exploration for research, teaching, and edge-oriented workflows.
Science
On-device sequence analysis and design exploration for research and productivity workflows.
Developer tools
A mobile hex viewer, byte editor, and binary inspection toolbox.
Collaboration
Universities, research groups, and companies hire Japality for AI systems, scientific software, and product engineering. Deliverables, fees, and data handling are set in writing.
How collaboration worksClear problem firstIntended users, success criteria, and constraints before feature lists.
Software you can runTools, reports, method notes, and handover—not an open-ended retainer by default.
Agreed file transferShare project files only after scope and terms are written down.
How a project starts
Tell us the problem and the outcome you need. We review fit and scope before any detailed data exchange or development work.
Organisation, objective, expected outputs, and approximate timeline—without confidential payloads.
Technical fit, inputs, risks, responsibilities, and whether we are the right partner.
Deliverables, schedule, confidentiality, IP, and commercial terms before work starts.
Contact
Send a short, non-confidential description. The same address covers commercial licensing, app support, and company enquiries. Do not send credentials or sensitive datasets in a first email.