Mark Gordon / Promethean Endeavors
Promethean Endeavors · Mark Gordon's application lab

Four production AI applications, built while learning a new way to build.

A former pharma-intelligence product executive's second act: building production AI applications hands-on, with a development practice designed around AI agents from the start.

Promethean Endeavors is my LLC and public workshop: part portfolio, part demo lab, part field notes for people building AI-native applications.

This site isn't a résumé. It's an argument about what's now possible — and an open invitation to collaborate, compare notes, and share best practices with people building the same way.

Speak
Try it — live, with your own words Talk the way you actually would after a meeting — or start from an example like "Coffee with Sarah Chen from Acme — follow up on the Q3 proposal by Friday" — and watch it become structured tasks, contacts, and notes in real time. Powered by Murmur, one of the four apps below. No sign-up, nothing saved.
4apps in production
120,000+lines of application source
850+test files
Dec 2025first line of code
№ 1

The Apps

Otis

Enterprise · Biotech Research
In short

An agentic AI platform for biotech research, in production and in daily professional use. Otis is a pseudonym; client-identifying details are blinded.

ExpandCollapse

It plans and executes autonomous research across SEC filings, clinical-trial registries, and news; writes citation-backed reports; and answers questions through an assistant that double-checks its own work.

Read more → ◆ Deeper walkthrough for invited guests

Murmur

Voice · Structured Capture
In short

Murmur is the missing input — speak a messy thought and it becomes structured records.

ExpandCollapse

Every CRM, task manager, and note system shares the same weakness: they're only as good as what people bother to type into them. Murmur turns messy thoughts into tasks with due dates, contacts matched to the ones you already have, and notes linked to the right project.

Read more → Try it live →

Diviner AI

Forecasting · Collective Intelligence
In short

Diviner AI asks whether collective forecasting discipline can be reproduced with LLM personas.

ExpandCollapse

Diviner started with a human expert crowd: independent forecasts from experienced drug developers, aggregated into one probability. Diviner AI asks whether that same discipline can be reproduced with LLM personas.

Read more → ◆ Deeper dive for invited guests

BG Watch

Personal · Built for Two
In short

A television tracker for exactly two users: my wife and me.

ExpandCollapse

What should we watch next? Did we ever finish that season? Where are we watching that — Netflix, Hulu, Prime? Ugggghhhhhh. A small app that answers the questions every couple asks each other — built with the same rigor as the bigger applications.

Read more →

Otis is the kind of system I used to expect a full team to build.

Earlier in my career, I led teams building applications of similar complexity. This time, with AI coding agents, strong tests, automated review, and a lot of process wrapped around the tools, I built and shipped the first production version myself.

The point is not that teams are obsolete. The point is that software work is changing quickly, and keeping up with all the changes is a big challenge.

№ 2 · The Practice

The apps are half the story. The other half is the practice I'm learning as I build them.

Isolated test-driven AI subagents that can't see each other's work. Parallel development across git worktrees, orchestrated like a small team. Automated product exploration that screenshots every user journey and critiques it. This page documents the whole practice — it's the one most CTOs ask about first.

Read: How I Work →