Independent R&D Lab · Established 2019 · Colorado

koderex.ai
Rev. 2026.09
Status: active

An independent R&D lab for software manufacturing — the engineering discipline of building and governing systems that produce software.

Operator
Travis Frisinger
Day job
Head of Agentic AI, 8th Light →
Output
Papers, talks, working code

§01
Abstract

Kode Rex is where I pressure-test ideas about how software gets made when the machines write most of it. The work is less about polished demos and more about the control problem underneath: governed autonomy, evaluation, context engineering, and what has to be true before anyone can trust a system that nobody reads line by line.

Experiments here become papers, talks, and a book. Anything that survives contact with a real product gets used at Void Works, the studio where the ideas have to actually ship something. What survives that goes into production on client systems, where being wrong is expensive — experiment, proof, production, and nothing skips a stage.

§02
Lines of
inquiry

Software manufacturing
A control architecture for building and maintaining software when nobody reads the code.
Governed autonomy
Which check enforces each rule, and what happens when it fails. A rule with no enforcing check is an opinion.
Evaluation systems
How you judge the judge, and how you know autonomous behaviour has not quietly regressed.
Context engineering
Not whether the agent has instructions, but whether it receives what it needs to decide reliably.

Fig. 1
Method

01

Take an idea

Usually one that will not leave me alone.

02

Build it for real

Not a demo. Something that has to keep running.

03

Break it

Find where the control actually fails, not where it looks fine.

04

Write down what held

Receipts and failures included. That becomes the paper.

Fig. 1 — the loop. Step 03 is the one most labs skip.

§03
Papers

Current · White paper · September 2026

Software Manufacturing

A control architecture for building and maintaining software when nobody reads the code.

Read the paper (PDF) →

Apr 2025

AI Coherence: a theory of utility in large language models

Mar 2025

2,500 ChatGPT conversations: a case study

2026

When Code Becomes FLUID, Where Does the Engineer Go? — book, in progress

§04
Talks

Upcoming · Talk · November 2026

When Software Builds Itself: How the Engineering Gets Encoded

Beyond Conf

Intent comes in as an issue and an autonomous pipeline moves it through design, tests first, build, review and merge. No human reads the diff. What makes that safe is not better code generation — it is everything wrapped around it. Plus the night it failed anyway.

2026

Self-Building Software

GOTO Accelerate Chicago — what it takes to trust a machine with merge rights

2025

The Coherence Trap

AI Engineer World’s Fair — why LLMs feel smart but are not thinking

2025

The Value of LLMs

LambdaConf — why ‘stupid’ machines are smarter than you think

2025

LLMOps: Taming Probabilistic Systems

DevOps Midwest

§05
Elsewhere

AI Buddy →

Essays and the papers

TDD Buddy →

A blog that writes and publishes itself

LinkedIn →

Where to reach me

GitHub →

The lab’s working code