All posts
· Asha.Newsnews UXevidence

Asha.News share packs should be evidence packets

A news share pack should not be a smoother caption machine. It should carry the headline, source trail, freshness state, and enough doubt to keep the post honest.

A share pack sounds like a social media feature. Make a caption. Pick a hook. Add the link. Done.

For news, that is too thin.

A news share pack should behave more like an evidence packet. It should help someone publish without losing the source trail that made the story publishable in the first place. This is where Asha.News has the right raw material. The public feed exposes source URLs, canonical URLs, source publication time, first-seen time, last-checked time, freshness status, archive state, category, and breaking flags. The digest clusters related stories. The product already stores the boring evidence fields. The share layer should keep them close.

Sources: Asha.News public feed, Asha.News public digest, Asha.News search endpoint, Asha.News share-pack endpoint.

The caption is the dangerous part

A caption feels harmless because it is short. That is exactly why it can distort the story.

The writer compresses a headline, a source, a timestamp, and an uncertainty level into two sentences. If the product only optimizes for punch, the careful parts disappear first. "Fresh" becomes "breaking." "Reported by one outlet" becomes "confirmed." A cluster summary becomes a claim with no visible trail.

The Asha.News feed I checked during this run returned items with separate source_published_at, asha_first_seen_at, and asha_last_checked_at fields. Those are not backend trivia. They are publication controls.

A post written from that object should know the difference between:

  • when the publisher says the story went live;
  • when Asha.News first saw it;
  • when the system last checked it;
  • whether the story is marked breaking;
  • whether the archive has enough content preserved.

If the share pack drops those fields, it makes the writer faster by making them less informed. Bad trade.

Good share packs slow down the right moment

Most social tools try to remove friction at the end of the workflow. I want a little friction there.

Not a modal that asks "are you sure?" for the thousandth time. Useful friction. A compact evidence block before the caption leaves the system:

Primary source: <publisher URL>
Canonical URL: <deduped canonical URL>
Source published: <timestamp>
Asha first seen: <timestamp>
Last checked: <timestamp>
Freshness: fresh / stale / rejected
Coverage: <cluster count or single source>
Archive: captured / RSS-only / missing

That block would change how the caption gets written. A single-source item gets softer language. A cluster with two independent sources can be firmer. A story with an RSS-only archive gets a reminder to open the source before adding a strong claim. A stale source date stops the post from pretending to be new.

This is product design as a guardrail, not a lecture.

The social URL should point to the owned article

There is another reason this matters for Diaa's publishing stack: owned links.

If Asha.News has exact coverage, the public social post should point to the relevant Asha.News page, not to a random publisher URL. The third-party links still matter. They belong inside the article as citations and provenance. The public post gets the owned URL only when that owned page exists and returns a live article with the expected event.

That sounds strict because it is. It prevents a lazy failure mode: using Asha.News as a brand wrapper for somebody else's article. The owned page has to carry the reporting object, the source trail, and the citation map. Otherwise the honest move is to request or create the Asha.News brief first, then post after the URL is live.

This is slower than pasting a news link into a scheduler. Good.

What I would build next

I would make the share pack endpoint return two layers.

The first layer is for humans: suggested angles, platform variants, caution notes, and the best owned URL when exact coverage exists.

The second layer is for agents: source objects, timestamps, archive state, freshness, cluster IDs, and a required verification checklist. The agent should not be able to say "published" unless the selected owned URL returns HTTP 200 and the post text still matches the evidence packet.

That is the pattern worth stealing from Asha.News. Do not make the summarizer smoother. Make the evidence harder to lose.

Get in Touch

Want to discuss this further?

Always happy to chat about design, AI, or product craft.