Perrin Research · Idea Foundry

Ideas in.
Evidence out.

We translate ideas into working prototypes, measure them for value, and ship the ones that earn it. Interactive ideas become applications. Static ideas become writing. AI agents run the loop.

The operating model

Every idea earns its way out

Nothing ships on opinion. Ideas become prototypes, prototypes generate evidence, and the evidence decides.

If the idea is interactive

it becomes an application

web apps · models · tools

If the idea is static

it becomes writing

essays · articles · books

Powered by AI agents

AI agents are the machinery of the foundry. They work up prototypes, run evaluations, and cycle ideas through the loop far faster than a traditional lab — so more ideas get a fair test, and only the valuable ones consume real time.

Now in the pipeline

Live from the foundry

8 prototypes tracked · 7 in flight · 1 shipped

PR-004▤ static · Blog

Leading with Attention

Lessons from AI software engineering translate directly into how technical teams should be led.

Shipped

▲ Live on Medium — recurring series

PR-000◇ interactive · Application

Perrin Research

A public site that explains the foundry operating model and pipeline can attract collaborators and validate the portfolio thesis.

Evaluating

▲ Live at dev.perrinresearch.com — production at perrinresearch.com

PR-002◇ interactive · Application

Waypoint

Portfolio surveillance focused on a sector or disease area surfaces where capital and science converge before the market does.

Evaluating

▲ Working prototype — daily internal use

Raw material

Where our ideas come from

Prototypes are only as good as the domain depth behind them. Ours comes from four places.

01

AI / ML

Machine learning and agentic systems — both the subject of our work and the machinery that runs the foundry.

02

Muscular Dystrophy

Duchenne and related conditions, from molecular models to exercise science and patient impact.

03

Drug Development

Translational insight connecting biological mechanism to therapeutic and economic outcomes.

04

Biophysics

Computational modeling of proteins, membranes, and molecular systems at atomistic scale.

The bet

Most organizations decide what to build, then build it. We build first — cheaply, with agents — and let measured value decide what deserves to exist. This site, like everything we make, is a prototype that earned its place.