About
Who builds this, and under what authority
Republic argues that authority and provenance should be resolvable rather than asserted. That standard applies first to this page — so the checkable part comes before the part you have to take on trust.
Authorship
Kyle Scott
Founder, Eldritch Labs. Architect of the Cornerstone AI Development Mesh. Technology leader across AI platforms, cloud architecture, and enterprise systems — Minneapolis / Elk River, MN.
Republic is what a year and a half deep in AI development and governance taught me — building agent platforms, watching where they hold and where they fail, and learning what it actually takes to trust autonomous work. The paper took weeks; the lessons behind it took the year and a half.
Republic is designed and built by Kyle Scott, with Claude (Anthropic) as a collaborator on research synthesis, drafting, and implementation under the author’s direction. All rulings are the author’s — recorded in the corpus’s decision history, not summarised here.
That disclosure is unusual and it is deliberate. A governance platform for AI-built software that quietly used AI to write its own case, without saying so, would be making its own argument against itself. The separation the system enforces between authorship and authorization is the same separation stated here: the drafting was collaborative; the rulings were not.
Authority is granted, never inferred from behavior, history, or kind. The corpus applies that to agents. This page applies it to its author.
The record
Selected work
The items below are stated biography — attested, not checkable from here. The one that can be inspected is Republic itself, and it is the one this site asks you to weigh.
Sole designer and builder of the Cornerstone AI Development Mesh
30+ services · 1M+ lines of original code
A governed multi-agent development platform with platform-native AI governance, built end to end. This is the prior work Republic stands on. The problem Republic addresses — that verification, not generation, is the binding constraint — is not a thesis arrived at from the outside. It is what building and running a multi-agent platform at that scale surfaces.
Six-plus years at Amazon Web Services
Sr. Cloud Application Architect · Technology Lead · Product Owner
AI modernization offerings that scaled to a $500M global focus; hyperscale platforms supporting more than a million nodes; and Amazon Bar Raiser.
That last one is worth pausing on, because it is not decoration here. The Bar Raiser program is one of the mechanisms Bryar and Carr document in Working Backwards — a structural check placed in the path of a decision, held by someone with authority to stop it, precisely so that the outcome does not depend on everyone involved trying hard. Republic’s central claim — good intentions don’t work; mechanisms do — is not a quotation picked up from a book. It is a doctrine served under, then built into software.
Engineering and platform leadership before AWS
Best Buy · Digi International · Thomson Reuters
Including an IoT platform supporting roughly a million concurrent devices. Regulated-environment experience across HIPAA, PCI, and banking — the settings where the cost of an unverified claim is not theoretical, and where structural controls are already the norm rather than an argument that has to be won.
Education
M.S., Software Engineering
University of Minnesota
B.S., Information Technology
Certifications
AWS Certified Solutions Architect – Associate
AWS Certified Cloud Practitioner
Certified Information Technology Architect (CITA)
Iasa Global
Supervised Machine Learning: Regression and Classification
DeepLearning.AI / Stanford, 2024
Core domains
- AI / agentic systems
- Governed multi-agent platforms
- LLM application architecture
- Cloud-native & hyperscale systems (AWS EKS, multi-cluster)
- Enterprise architecture & modernization
- DevSecOps / SRE / FinOps
- Regulated environments (HIPAA, PCI, banking)
Ownership and stewardship
Republic and Eldritch Labs are distinct
These facts are close enough to collapse by accident, so they are stated separately and precisely.
Republic
Owned by Kyle Scott personally. Developed in the open and stewarded by Eldritch Labs.
It is not an Eldritch Labs product. The distinction is corroborated in the record rather than only asserted here: the proposed outbound license, DEC-0002, is granted personally by the owner.
Eldritch Labs
The trade name of ABG Technology Consulting LLC, solely owned by Kyle Scott.
Eldritch Labs stewards Republic’s development in the open. Stewardship is custodial: it does not confer ownership, and it is not a funding relationship dressed up as one.
eldritchlabs.net ↗Cornerstone AI Development Mesh
Owned by ABG Technology Consulting LLC. A service mesh for coordinating AI agents across independent repositories — 30+ services, architected and built by Kyle Scott.
Cornerstone is not part of Republic, and Republic is not built on it. It is named here because it is the prior work — the platform whose operation surfaced the problem Republic exists to solve.
How to evaluate this
Read the corpus, not the summary
Everything above this line is a claim about a person. Useful context, and none of it is the reason to take Republic seriously.
The strongest thing that can be said for this project is not on this site at all. It is that the substrate is public, the evidence rows are committed, the conformance workflow runs on every change, and the standing queries are re-runnable by anyone against any commit.
The whitepaper states the criteria by which its thesis could be proven wrong, and the system emits evidence of its own performance — including evidence of failure — as a side effect of operating. That is the claim worth testing, and the means of testing it are public.
Contact
Review, refutation, and enquiries
Republic invites external review, and it has already benefited from it — the composition defect described on the current-state page was found by an outside reader, and is published rather than quietly patched.
Machine review was the first gauntlet. The one that matters is human. If you build agent platforms, run governance programs, or review systems work professionally, findings are wanted — and every one gets the same treatment the machine findings did: applied, declined with reasons, or tracked openly, on the public record and credited to you if you want it.
A technical finding
Raise it against the record itself, where it becomes part of it. This is the preferred route — a finding filed in the open is already halfway to being dispositioned in the open.
github.com/tecthulhu/republic/issues ↗Anything else
Enquiries, correspondence, or a finding you would rather not open in public first.
republic@eldritchlabs.net