Content Operations Control Center
A U.S. healthcare content team had an automated pipeline running — research to rewrite to narration to video to multi-channel publishing. What they didn't have was a way to tell if any of it was actually working without opening five different systems and checking each one by hand. This dashboard is that missing layer.
- ROLE
- Dashboard architecture, data-authenticated QA, SEO/analytics diagnostics
- TOOLS
- ChatGPT · Dashboard architecture · Analytics diagnostics
An automated pipeline nobody can see into isn't actually safe to run.
The underlying workflow was doing real work — pulling research, rewriting it, generating narration and video, and pushing finished pieces out to social channels. None of that was visible in one place. Confirming whether a piece had actually rendered, whether a render had failed, or which platforms already had content meant logging into the workflow tool, then the database, then storage, then each channel separately. A failed run stayed invisible until someone happened to notice content was missing.
Six views into a pipeline that used to be a black box.
Built on top of the existing automation rather than replacing any of it, the dashboard reads directly from the automation's own data as it's produced, and turns that data into views a nontechnical operations team can check on its own, without touching the pipeline itself.
High-level status across the whole pipeline: what's running, what's queued, and whether today looks normal.
Every generated article in one searchable place, instead of scattered across workflow runs and storage folders.
One article, with its narration, video, and captions shown together instead of spread across three systems.
A record of every media render, so a failed or stalled video job is visible without checking the render service directly.
Channel-by-channel publishing status and the exact caption that went out with each post, per platform.
Workflow failures surface here immediately, instead of staying silent until someone notices content is missing.
A dashboard is only trustworthy once it's been checked against reality.
Before this could be treated as the source of truth, its own numbers had to be verified against the real system — authenticated QA run directly against live production data, specifically to separate a genuine dashboard bug from an error that already existed further upstream in the pipeline. Only once that distinction was clear could the operations team trust what they were seeing instead of double-checking it by hand every time.
AI as technical investigator, not just content generator.
The same engagement ran technical SEO and analytics diagnostics on the client's website in parallel — auditing the site's analytics and search-console configuration, validating canonical URLs, and tracing a third-party SEO script that was quietly interfering with page-head rendering before replacing it with a more reliable implementation.
Running automation nobody can actually see into?
Tell us what's running blind today, and we'll map what a practical implementation could look like.