The LogicLegion Papers
Three companion papers on building agentic software that enterprises can actually ship: why the discipline is different, how to stand it up, and what becomes defensible once you do.
Why is governed agentic delivery different?
Classic engineering principles relied on determinism and the data/code boundary. Agentic systems remove both. Harness engineering re-derives each principle as explicit controls outside the model — and the one invariant underneath is verification capacity.
Theory paper · ~29 pp.
What do we do Monday morning?
The theory aimed at execution: a phased, 90-day path to stand up governed, shippable agents. Owner first, eval harness before the agent, gates and evidence around it — the dependency order that separates a shippable system from an impressive demo.
Practice playbook · org + technical tracks
What is actually worth owning?
For most teams, rent the frontier and route. A durable moat exists only where proprietary data, a regulated workflow, and deployment locus reinforce one another. The asset to own is not the model — it is the flywheel that re-renders capability as the ground moves.
Strategy brief · placement & moats
The set reads as one argument across three altitudes. The theory paper establishes the invariant — verification capacity — and why it outlives any model. The playbook turns that invariant into an operating order a team can follow this quarter. The strategy brief answers the ownership question the other two raise: most teams should rent and route; only the rare team with proprietary data, a regulated workflow, and deployment locus should build the flywheel.
Read the theory to understand the why, the playbook to execute, and the strategy brief to decide what to bet on.
Enter your email to unlock the PDFs. One email, all three papers.