An AI-Driven Editorial Tool That Cut Turnaround Time by 60%
My Role
Solo UX Designer, led end-to-end product design, from understanding workflows to designing the complete tool experience.
Deliverables
-
UX architecture
-
Flows for editorial and design workflows
-
Wireframes and UI design
-
Prototypes for internal reviews
-
Design documentation and release support
Team
-
2 Developers
-
1 Machine Learning Engineer
-
Editorial Director
-
Myself (Solo UX Designer)
The Problem
The newsroom team, spread across six languages and made up of editors and designers was struggling with a fragmented, inefficient workflow. Core editorial tasks like writing headlines, summarizing articles, sourcing images, and publishing content were not only repetitive but also handled through disconnected tools, slowing down the entire process.
Key challenges included:
-
No centralized view of incoming news from over ten publishing partners, making it difficult to prioritize and coordinate coverage
-
Editors and designers worked in parallel but siloed workflows, resulting in slow communication and delayed approvals
-
Tools used for headline writing, summarizing, image selection, and publishing were entirely separate, causing friction and delays—especially during time-sensitive breaking news
-
Limited access to relevant visuals during article creation forced teams to rely on manual image sourcing through external platforms
-
There was no system to auto-generate article copy or suggest images based on article content, requiring extra manual effort and slowing down production
It became clear that the team needed a unified tool—one that could integrate AI to generate content, recommend visuals, and bring editorial and design efforts into a shared, streamlined workflow.
The Goal
The objective was to design an internal tool that could streamline article creation across editorial and design teams while reducing dependency on multiple disconnected systems.
Specifically, the tool needed to:
-
Ingest news content from over ten RSS feeds
-
Auto-generate headlines, summaries, and image prompts using AI, with minimal need for manual editing
-
Enable editors and designers to work within a shared interface, avoiding tool-switching
-
Provide real-time visibility into article progress and team assignments
-
Support scheduling, resurfacing, and full control over the article lifecycle
-
Improve publishing speed and consistency without compromising editorial oversight
Listening &
Learning
To design a system that fits real needs, I worked closely with the people using it daily:
-
Stakeholder Interviews — editors, designers, team leads, and publishing managers
-
Live Observations — shadowed editors and designers to identify behavior patterns and friction points
-
Surveys — captured gaps in current workflows and tool expectations
-
Feedback Loops — shared iterative prototypes and refined the UX with direct feedback
Key Insights:
-
Editors wanted automation, but also full control
-
Designers needed faster image generation with fewer clicks
-
Both groups needed better task tracking and less mental overhead
-
Most users preferred working top-down — from text to image to publish
DISCLAIMER:
(a) Displayed brand logos and brand names in the portfolio and CV, represent past client work and are showcased solely for portfolio and resume validation. They do not imply direct endorsement or affiliation. (b) Certain visuals used for the portfolio include copyrighted stock images and videos. All the visual media which includes images, GIFs, videos and texts are strictly for portfolio display and must not be used or reproduced without proper authorization from the copyright owners.
The Solution
A unified tool with three core work areas, Article Text, Cover Image, and Publishing, designed to balance automation with clarity and control.
1. Homepage: New Articles Queue
The homepage serves as the central hub where all new articles are displayed in a list view. Each article entry includes essential details:
-
Time the article was received
-
Category
-
Assigned editor and designer
-
Priority level
-
Breaking News status
At the top of the page, language tabs allow users to quickly switch to their specific language or view all articles across all languages. This makes it easy to see what has been published in other languages and duplicate articles when needed for multilingual coverage.
The filtering system provides options to refine the list by time duration, category, language, and an “Assigned to Me” toggle, which can also be activated instantly using the Q shortcut. A global search bar allows searching by headline, publisher, or keywords, using the shortcut Cmd/Ctrl + K.
This design ensures that at a glance, editors and designers can assess priorities, identify unassigned articles, and act quickly, especially when dealing with breaking news.
2. Editing Copy
The Article Text tab is where editors refine the written content before publication. To save time, the system uses AI to automatically generate the headline and description in the background as soon as an article is created. This means that when an editor opens the page, the suggested copy is already in place and ready for review.
Editors can:
-
Edit the AI-generated text directly
-
Regenerate new suggestions if the initial version is not suitable
-
Approve the copy by clicking Confirm Text, marking it as editorially complete
Beyond text editing, this section allows editors to adjust key metadata, such as:
-
Changing the article’s category or language
-
Updating its priority level
-
Tagging it as Breaking News or a Featured article
For quick workflow management, editors can assign or unassign themselves to the article directly from this view, either by clicking the Assign button or using the Space keyboard shortcut.
A live progress indicator ensures clarity on what remains to be done. For example, if mandatory fields are incomplete, a message like “2 fields need attention” appears at the top of the section.
3. Article Images
The Cover Image tab is where designers select and prepare visuals to accompany an article. Each article can have up to four images, which are shown in rotation whenever the article resurfaces, ensuring visual variety over time.
Designers have three ways to add images:
-
Generate new visuals using AI models, including Google Imagen 3 or Stable Diffusion
-
Select from the integrated internal image library for quick reuse of existing assets
-
Upload custom images directly into the tool
To speed up the process, an image prompt is automatically generated based on the article content. Designers can use it as-is or refine it to produce more specific results. When generating images, each prompt produces four variations, allowing quick comparison and selection.
Image generation runs asynchronously, so designers can continue working while the system creates visuals in the background. A countdown timer shows exactly when results will be ready, helping maintain workflow efficiency.
4. Publishing
The Publishing tab is the final step before an article goes live. It provides a complete overview of the article’s text, images, tags, and metadata, ensuring that editors and designers can verify every detail before publishing.
From this view, users can:
-
Publish Now for immediate release
-
Schedule a specific publish date and time using an integrated calendar
-
Set resurfacing rates to control how often the article reappears in feeds over time
-
Review the full article layout as it will appear to readers, including headline, description, images, and tags
This consolidated view eliminates the need to switch between tools or preview environments, allowing quick adjustments and confident publishing decisions. By combining scheduling, lifecycle controls, and a live preview in one place, the publishing process becomes faster, more reliable, and less error-prone.
The Outcome
The rollout of the new tool transformed the way the newsroom operated, delivering measurable efficiency gains and improved collaboration.
Key Results:
-
60% faster article completion for editors, driven by AI-assisted copy generation, streamlined workflows, and reduced manual input
-
45% faster image preparation for designers, thanks to integrated AI generation and direct access to the internal image library
-
25% reduction in publishing errors, achieved through confirmation steps, clear progress indicators, and real-time status updates
-
Faster onboarding for new editors, as the AI-generated starting point for headlines, descriptions, and prompts helped them get up to speed quickly
-
Elimination of tool-switching, allowing editors and designers to work entirely within a single, integrated platform
-
Full transparency on article status, enabling better coordination between editors and designers and faster handoffs
-
Consistent content quality, as AI ensured that initial drafts met editorial standards across languages and categories
Beyond the numbers, the tool became the single source of truth for all editorial work — consolidating content creation, image workflows, and publishing into one cohesive system. It empowered teams to work faster without sacrificing accuracy or control, making it easier to respond to breaking news and manage high-volume publishing schedules.
