THE TATVA · CONTENT INTELLIGENCE

A newsroom loop built around a neutral voice.

I worked directly with The Tatva's founder to translate an established, plain-spoken news format into a content platform that could discover, regenerate and learn from every story.

Hand-drawn window scene of a person reading the morning newspaper with coffee

Role: Product strategy, system definition & UX direction
Product: Editorial content-generation platform
Core principle: Preserve the publication's neutral voice

01 · THE STARTING POINT

The product had to learn the newsroom—not replace it.

The Tatva had already built its audience around direct, non-opinionated news. The opportunity was to make that editorial method repeatable across a much larger stream of incoming stories without flattening the tone that made it recognisable.

The system therefore followed the existing generation process: source material came in, context was retained, the story was rewritten in The Tatva's voice, and performance data informed what the team looked at next.

02 · DEFINING THE VOICE

“Neutral” had to become a system the product could follow.

I studied the publication's archive with the founder, separated repeatable editorial conventions from subject matter, and translated them into generation and review criteria.

01Read the archiveHeadlines, sentence shape, context and endings

→

02Name the attributesDirect · factual · concise · non-performative

→

03Build the checksSource fidelity, unsupported tone and human review

→

04Connect the loopGenerate, publish, analyse and improve

The defining decision: ground generation in the publication's own archive and editorial conventions—not in a generic “news voice.”

Hand-drawn window scene of a person reading on a phone while watering a balcony plant

Hand-drawn window scene of a person pinning a news clipping

Hand-drawn arched window scene of a person reading on a tablet

Hand-drawn window scene of a person reading at a kitchen counter

03 · THE PRODUCT SURFACE

One working environment for the editorial team.

The interface brought user management, content, the story library and trending signals into one operational layer. The full design file is embedded below so the screens remain evidence rather than decoration.

04 · THE SELF-FEEDING SYSTEM

Three pipelines, one continuous editorial loop.

GENERATE + REGENERATE

Turn source material into Tatva-ready copy.

A content-regeneration pipeline accepted the source news, retained the factual context and produced a concise draft aligned with the publication's existing archive, platform patterns and neutral voice.

DISCOVER

Bring the wider news cycle into view.

Competitor and publisher RSS feeds formed a discovery stream, helping the team see what was emerging across the category and decide what deserved attention—without turning the platform into an opinion engine.

ANALYSE

Move from page metrics to useful questions.

Detailed performance analytics made it possible to compare pages, recognise patterns and inspect what was working. A conversational AI layer let the team question the data directly instead of navigating reports alone.

discover→generate→publish→learn↺

05 · THE PRODUCT IDEA

Scale the process.
Keep the judgement human.

The platform connected discovery, drafting and performance into one clear loop. AI accelerated the work around the newsroom; it did not become the publication's point of view.