The problem
S.A. Lifeline Foundation came to us with a site that had years of content and no time to fix it. Blog posts sat in a dated layout. Titles and meta descriptions hadn't been touched in years. The crawl output was full of broken internal links, duplicate tags, and missing alt text — the kind of backlog that never gets prioritized because no single item on it looks urgent. And years of their work sat locked in podcast audio, invisible to search.
What we did
We worked through the backlog with the same pattern we run on every engagement: AI does the repetitive volume, a person makes the judgment calls on what ships.
- Technical SEO. Triaged the crawl by type and severity — broken internal links, duplicate and missing tags, schema errors, redirects, missing alt text — so cleanup took hours instead of weeks. Titles and meta descriptions were rewritten in batches, each reviewed before it went live.
- A podcast-to-web pipeline. We built a WordPress plugin that syncs new episodes from the feed, transcribes them, uses an LLM to label and name speakers, and publishes a rich, structured page per episode — turning 104 audio-only episodes into indexable pages, cross-linked from the existing blog.
- An image service. After a paid third-party optimizer failed and burned their credits, we built our own — a small Node/Fastify/sharp service on Fly.io plus a WordPress plugin — to compress and modernize the whole image library, with an AI upscaler for the pixelated ones. (That work became the seed of a product of our own.)
- Blog redesign. We built one on-brand template, then had AI apply it across the highest-traffic back catalog while a person spot-checked — days, not months.
The outcome
The backlog got worked, the image library got fixed, and years of audio became pages people can actually find. None of it was AI working alone: it handled the repetitive volume, and a person made the calls on what shipped. That's the shape of every engagement we run.