// case studies
Six brands, six content engines. Same underlying stack (Mastra + Vertex AI + DataForSEO + Ghost/WordPress + Late), very different voices. Here's what each one ships and what's hardest about it — no fabricated metrics, just the actual work.
// what the engine ships
STR cohosting brand I co-run with Alan. Engine ships ~3 blog posts/week + 5 socials/week via Late, hyper-local content targeting the markets we operate in.
// hardest part
Hardest part: local content sounds generic the moment you let the LLM lead. We use neighborhood research as ground truth and the model paraphrases — not the other way around.
// what the engine ships
Our AI assessment / consulting studio. Engine ships SEO-led pillar pieces and case studies, plus LinkedIn distribution. Different voice file from this site — sharper, more B2B.
// hardest part
Hardest part: keeping voice consistent across both my and Alan's posts on the same blog. Solved with per-author voice files plus an editorial gate.
// what the engine ships
Alan's fractional CTO + AI-for-trades brand. Engine ships practitioner-tone blog content and weekly LinkedIn, leaning hard on operator stories from real engagements.
// hardest part
Hardest part: this audience smells AI copy at 30 yards. Engine drafts; Alan rewrites in his voice. The engine's job is to reduce the blank page, not replace him.
// what the engine ships
Done-for-you sites for local service businesses. Engine ships geo-modified informational content per client niche (plumbers, HVAC, etc.) and a refresh sweep keeps the calendar healthy.
// hardest part
Hardest part: scaling per-client voice without a separate workspace per client. We use a single Mastra workspace with brand IDs and pgVector RAG scoped by brand.
// what the engine ships
Proof-of-work + repair tracking for service pros. Engine ships product-led SEO and customer-story posts pulled from real fix tickets.
// hardest part
Hardest part: turning structured ticket data into stories without leaking PII. We strip and synthesize at the ingestion step before anything touches an LLM.
// what the engine ships
STR supply automation for cohosts and managers. Engine ships operator-focused content for property managers and a weekly digest of restock-pattern insights.
// hardest part
Hardest part: this audience is small and technical. The engine produces useful drafts but I edit hand-tight before publishing. It's the closest thing to my own blog in tone.