How the applications, shared services, CI/CD, and AI model routing fit together
These aren't four independent applications. They share a common engineering platform for AI, security, deployment and operations - and this map answers one question: how does it all fit together?
Applications - Tourney, Runway and GetSeen (plus this portfolio site) are the products. Platform - everything else on the map is what they run on: the five-gate CI/CD pipeline across the top, Control Tower as the one mandatory checkpoint every deploy and promotion passes through (and the AIOps loop that investigates a failed one), the shared modules beneath the applications (login security, AI tool loop, AI worker boundary, grounded AI answers), and the three AI backends chosen per workload on cost, performance and data sovereignty.
Click any box for the mechanism behind it — including the three numbered circles, which are the audit checks.
Solid lines are automatic and live. The dashed line is the one manual step (promote). The dotted boundary is local start/stop only. The accent-colored rail is the audit trail; ③ on Runway is a fourth kind of check — the app reruns its own math. Click anything for detail.
Nothing selected yet — click a box or a numbered circle above.
Line meaning
Automatic — deploys, imports, live calls
Manual — a person clicks (promote)
Local control — start/stop, never a gate
Would fit — same primitives, not migrated
Audit trail — append-only records along the flow
Auditing, at three altitudes
Pipeline — GitHub's own run history, per commit
Deploy — Control Tower's append-only promotion log