Web development
Production web apps, not prototypes. React and TypeScript on the front, a backend that holds up when traffic does on the back.
Apps · dashboards · APIs
We ship the hard parts.
Larynx™ is a New York software studio. We design and build web apps, mobile apps, AI and ML systems, and the cloud architecture that holds it all up — then hand them back to you, working, in production, documented.
NYC
Where we build
4
Disciplines, one team
Senior
No juniors, no handoffs
MedSynth™
Flagship · patent-pending
What we do
Four disciplines, one accountable team.
No account managers, no offshore handoff. The people who scope your project are the people who build it — across the whole stack, from the interface to the infrastructure.
Mobile apps
iOS and Android that feel native, shipped to the stores — not just to TestFlight. Cross-platform where it's smart, fully native where it counts.
iOS · Android · cross-platform
AI & ML
Retrieval, agents, fine-tuning, and evals — wired into your product with the guardrails and monitoring to run them in production.
Agents · retrieval · evals
Cloud architecture
The infrastructure underneath: containerized services, autoscaling, and deploy pipelines that don't page you at three in the morning.
Infra · pipelines · scale
Flagship project · medical AI · patent-pending
MedSynth™: drug discovery you can stand inside.
MedSynth™ is a spatial, agentic drug-discovery platform. It builds a three-dimensional representation of a biological target, proposes and re-scores candidate molecules with agentic AI, and lets scientists review the design session in augmented reality — on Apple Vision Pro, with a desktop fallback. It learns from federated clinical and real-world data without moving patient records, and plugs into the docking and free-energy engines labs already run.
View the pitch deck →Interoperates with Schrödinger, OpenEye Orion, MOE, and AutoDock.
Hover to play
Spatial review
Design candidates in the binding site in 3D on visionOS, with a 2D desktop fallback.
Agentic scoring
Multi-objective scoring for potency, toxicity, and synthesizability, run by an agent.
Federated learning
Re-weight scores against real-world outcomes without patient data leaving the custodian.
Enterprise interop
Re-rank externally generated candidate sets from the screening platforms you trust.
01 · Spatial review on visionOS
02 · Multi-objective scoring
03 · Federated, privacy-preserving
Who builds it
A small senior team that ships.
Larynx™ is deliberately small. Two people who have built and shipped real systems own every engagement directly — strategy, architecture, and the code itself.
AG
EL
Start a project
Tell us what you're building.
Send a paragraph about the product and where it's stuck. We'll tell you whether we're the right team and what the first month would look like — no sales call required.
info@thelarynx.com