La Diaria: TIA, a one-button title assistant that put stories back in search
One button in the CMS, three editable meta titles — adoption spread by word of mouth, then the tool lost its owner.
- How do you fit AI inside an existing publishing workflow?
- What drives adoption?
- What happens when a tool loses its owner?
Problem
Journalists at La Diaria were not writing meta titles, the short titles that search engines and Google Discover display. It was an extra step nobody took at the end of a long writing day, so articles were not appearing on Google Discover or in search, keeping content confined to the existing subscriber community. Reach beyond subscribers matters commercially as subscriptions account for more than 80% of the paper's income.
Solution
TIA ("Titulation by IA") adds one button inside the Content Management System (CMS) publishing flow. Before hitting publish, the journalist presses it and receives three editable meta-title suggestions. Human review is built into the design, and access is permission-controlled. The tool was calibrated with the paper's own articles paired with correct meta titles. The internal IT team did the technical work, its first AI project, across roughly three months. The tool was deliberately fitted into the existing workflow as a tiny step rather than built as a standalone product.
Results
Adoption grew by word of mouth, from three initial editors to all web editors and journalists, with colleagues requesting access. The newsroom regards that organic demand as the strongest proof of success. It worked best on short, fast-turnover content like breaking news, sports, justice and crime, where reporters said it produced more concise titles in fewer words. It performs poorly on long articles, such as six-page science pieces, which the design never considered. There is no analytics evidence that it improved long-term discoverability. Impact measurement stayed informal; the users' relief at not having to invent another title was considered sufficient, since that was the original motivation.
Insights
Eight to 10 journalists used it, mostly editors across sports, environment, justice, politics and web. One use that nobody designed for involved journalists feeding the suggestions back into the article's actual headline. The harder lesson is how the tool would be used afterwards. It is not currently working, and the team now brings titles to the staff member who built it, who runs them through an early prototype manually. Like Núbel (theme 4), a working prototype needs someone checking whether it still works, needs upgrades, or has stopped because a payment lapsed. Without that, the tool depends on individuals and never becomes institutional.
Cost
A monthly AI subscription of USD 190. The indirect cost was roughly three months of the internal IT team's time on top of normal duties.
Replicability
A one-button, human-supervised assist embedded and calibrated with a handful of house-style examples. It needs a CMS the newsroom can modify, some internal IT capacity, and style examples built from its own pages. It also needs to be designed for long-form content from the start.