Kasey Swails

My product, built and owned by me, with Sutcliffe Mechanical as the flagship customer using it on the shop floor every day. Six versions shipped on a continuous customer-feedback loop, with tenant isolation ready for additional orgs.

Daily shop-floor use in production · 2+ yrs running a real shop · 6 versions shipped on customer feedback · 90+ tests passing

A live platform at app.solvedtheproblem.com where new verticals assemble from existing primitives instead of starting from scratch. The billing rails are built on Stripe Connect, and the strongest pieces (the Rodeo vertical and the cross-tenant Governance/L10 layer) are now being extracted into their own standalone SaaS products.

5 verticals on one platform · 32 reusable platform modules · 2+ yrs live in production · Pre-revenue Stripe billing built, not yet earning

A local-first AI coding platform whose data never leaves the developer's machine: self-hosted, multi-model LLM orchestration for proprietary codebases that can't be shipped to a cloud vendor. It's my production multiplier: how I ship whole systems solo, faster and safer than teams renting per-seat cloud AI. The intended model is a one-time, org-bound license rather than a subscription. Nothing has been sold: this is a single-operator platform I run myself, and the licensing is built, not proven.

6 safety gates on AI-written code · Local LLMs, cloud optional · 88 subsystem tests · 7 major subsystems

A trusted, long-running engagement (~1.5 years of maintained production software). The schema-driven framework turns each new module into declarative config instead of bespoke code: proof of sustained reliability, not just a greenfield demo.

639 tests · zero failures · 1.5+ yrs continuous production · 6 handoff runbooks shipped

Self-hosted LLM inference versus cloud APIs: a build-versus-buy evaluation

kaseyswails.com/writing/self-hosted-llm-build-vs-buy

Criteria defined before scoring, benchmarks run on a two-tier GPU lab, a scored options table with sensitivity analysis, a total-cost-of-ownership model including the operator cost most comparisons omit, and a conditional recommendation.

Letting AI touch regulated systems

kaseyswails.com/writing/ai-regulated-systems

The five objections stakeholders raise to AI adoption, the specific architectural control that answers each, and how identity, integration, data, and observability interrelate.