Paper I is out: The Human Layer (DOI: 10.5281/zenodo.19119699). It is the foundation of everything Timer will build, so this month's entry is about what it says and why we published it before selling anything.
The claim is simple to state and unfashionable to hold: the most critical layer in enterprise AI is not artificial. It is the human layer: the judgment, context and accountability that sit between an AI system's outputs and the consequences of acting on them. The industry is currently spending historic amounts of capital trying to remove that layer. The evidence says this is backwards.
The pattern in the field data is consistent: organizations that design AI to amplify human capability outperform those pursuing full automation, not marginally but by multiples. Companies investing in human-AI collaboration grew faster while expanding their workforces. Replacement optimizes for cost; amplification optimizes for value. These are different engineering decisions, and they produce different companies.
Why publish this as a paper instead of a pitch deck? Because a pitch deck asks you to believe us, and a paper invites you to check us. Every claim in the series carries its citations. If we are wrong, we are wrong on the record, falsifiably, with a DOI attached. We think a company that wants to hold an organization's memory should be willing to be held to its own.
Publishing first also imposed a useful discipline internally. Writing "the human layer must be architected, not asserted" is easy; it commits you to actually architecting it, which is Paper II's job and, more to the point, our product roadmap's job. The papers are not marketing for the product. The product is the papers, made operational.
Meanwhile the company itself stays small and deliberate: invite-only deployments, founder-led onboarding, one workflow at a time. Slow is a feature this year.
Four papers to go.