AGENTS · BREAKING NEWS AGENT

Request access

Turn breaking signals into published stories, automatically.

AI news monitoring API with automated multi-platform publishing.

An agent you wire into your newsroom and CMS, not a tool your editors leave to go use. It watches 200+ sources across broadcast, video, text, and social, surfaces what is rising before the window closes, drafts in your editorial voice, ships the routine stories on auto, and routes the judgment calls to your editors with the draft already done.

200+
Sources watched 24/7
Platform-native cuts
10+
Languages, localized
1 API
One agent in your stack

WHAT THE AGENT DOES

  • Monitors 200+ broadcast, video, text, and social sources
  • Surfaces what is rising, scored and ranked, not editor bias
  • Drafts the article and captions in your editorial voice
  • Cuts a platform-native version for every destination
  • Routes routine stories to auto, judgment calls to editors
  • Publishes to YouTube, TikTok, Reels, X, LinkedIn, your apps

TRUSTED BY PRODUCT TEAMS SHIPPING VIDEO AT SCALE

Customer logoCustomer logoCustomer logoCustomer logoCustomer logoCustomer logo

WHAT YOU GET

A newsroom that covers everything, with the team you already have.

Point the agent at your sources and your CMS. It does the watching, scoring, drafting, cutting, and publishing, so your editors spend their time on the stories that actually need a human.

Nothing missed

200+ sources watched around the clock across broadcast, video, text wires, and social, so the story that lifts your numbers does not slip past at 2 AM.

Drafted in your voice

The article, captions, and script come back in your editorial voice and translated into 10+ languages, as seven platform-native cuts ready to publish.

You keep the judgment

Routine stories publish on auto; anything sensitive, ambiguous, or angle-driven routes to your editors with the draft already done. Approve once and every cut ships.

Do more with less

How a six-person digital newsroom chases a +40% DAU goal.

A four-panel story: the impossible goal, the view count Eva missed, the two-pipe Agent, and the +40% line crossed in Q2 — same headcount.

Panel 1 — The goal

"Q3 goal: +40% DAU. Headcount: six."

Eva sits at her desk in front of a wall chart with a tall plus-40-percent DAU target line. Six empty chair silhouettes line the wall behind her. 200+ sources to watch. Six editors to do it.

Panel 2 — The miss

"Meanwhile: 12M views elsewhere."

Editorial is heads-down on the morning beat at 11 AM. By 1 PM, a competitor's TikTok shows 12M views on the story that broke at 6 AM on X — the feed no one was watching while the team was in the meeting.

Panel 3 — The shift

"Agent ships 60%. You edit the 40."

Two pipes flow out of the Agent: a purple Auto-publish pipe shipping to YouTube, TikTok, Reels, and X; a green To-editor pipe with the judgment calls pre-drafted in Eva's voice. Eva edits the angle, not the boilerplate.

Panel 4 — Hit Q3 in Q2

"Coffee still hot."

The same growth chart from panel 1 reappears. The line crosses the +40% DAU target ahead of schedule. Sam, Eva's teammate, asks, "Wait — we already hit the number?" Eva: "Feed stayed fed. Same six of us."

IN PRODUCTION

What it looks like in your newsroom.

A real 15 seconds. Tuesday, 2:14:00 PM. Source to multi-platform publish — editor hasn't started writing yet, the article runs anyway.

A real 15 seconds. Tuesday, 2:14:00 PM.

Source → multi-platform · 15 sec
2:14:00
Three sources fire.
Reuters wire + YouTube live + X post on the same story. Cross-confirmed.
2:14:03
Score: 0.91 fit / 0.88 locality / business / neutral / 0.94 priority.
All 5 dimensions clear. Auto-route eligible.
2:14:08
Composed + routed.
Article + 4 social cuts in your voice. Auto-publish triggered.
2:14:15
Live on 5 platforms.
YouTube · TikTok · X · LinkedIn · Web. 5 cuts, 4 aspect ratios, 2 languages.

Editor hasn't started writing yet. The article runs anyway.

THE PAIN

You can't watch 200 sources. So you watch 20. And miss 180.

DAU goals don't care about your headcount. The story that lifts the line broke on a creator's X feed at 2 AM, in a podcast no editor follows, on hour 6 of a wire nobody had time to read.

Before

  • No team can watch 200 sources continuously. So you watch 20. The story that would've lifted DAU broke on the 180 you weren't watching.
  • Source monitoring is one person watching feeds. Bias by editor shift, not by signal.
  • Breaking stories miss the social window because the clip queue is hours long.
  • One producer per platform-native cut. Three platforms means three producers.
  • Editor reviews every clip, even the routine ones. Bottleneck on judgment calls.
  • Translation and localization are separate vendors. Lag of hours, not seconds.
  • "Should we post this?" is decided ad-hoc. No consistent editorial rules.

With Breaking News Agent

  • 200+ sources watched 24/7. Unbiased — the story score surfaces what is rising, not what an editor happens to follow on X.
  • Every source type covered: broadcast, digital video, text wires, social timelines, RSS, custom feeds.
  • Breaking stories surface within seconds. Composed in your editorial voice before you finish reading.
  • One story becomes 7 platform-native cuts automatically. Aspect, length, captions, brand graphics.
  • The score routes routine stories to auto-publish. Editors only see the ones that need a judgment call.
  • Built-in translation. 10+ languages, localized angles per region.
  • Routing rules are predictable: sentiment + topic category + source tier.

WHAT IT LOOKS LIKE IN YOUR NEWSROOM

Sources in. Cuts out.
Across every channel.

Two pipes: monitor runs against 200+ sources continuously. Publish runs per scored story. Both share the same FLASH scoring engine and voice model.

MONITOR200+ sources, 24/7
200+ sources
Dedup +
cluster
FLASH
scored
Routed:
auto or editor
scores ↑one shared FLASH + voice model↓ composes
PUBLISHevery scored story
Composed in
your voice
7 platform cuts
Editor
approves 40%
Published to
every channel
02:1417:46

Source-to-published median ~90 seconds. Live broadcast cuts ship in ~12 seconds with mid-event Clipping. Every webhook is HMAC-signed and replay-safe.

FLASH. FIVE DIMENSIONS PER STORY

Every story is scored on five axes

FLASH is the dimensions a working editor evaluates instinctively in the first 30 seconds of seeing a story. FLASH makes that evaluation explicit, scoreable, and consistent across hundreds of stories a day.

F

Fit

Matches your editorial voice and brand guidelines. Customer-tuned per account during onboarding.

L

Locality

Geographic relevance to your audience. Region, city, demographic. Drives per-region routing.

A

Angle

Suggested POV or framing. Business, political, human-interest, regional. Pre-drafted for editor review.

S

Sentiment

Positive, neutral, or negative reading. Triggers editorial routing when polarizing or crisis-adjacent.

H

Headline-worthy

Priority and breaking-ness. Drives publish urgency and platform ordering. Highest H scores ship first.

THE SPLIT

Agent ships 60%. You edit the 40 that needs a human call.

Tier-1 sources, neutral sentiment, FLASH-cleared stories auto-publish. Polarizing sentiment, crisis topics, and low-confidence sourcing route to editorial review with the cut already pre-drafted. Editor edits the angle, not the boilerplate. Approve once — 7 platform cuts ship.

FIVE STEPS, SOURCE TO PUBLISH

Monitor. Score. Compose. Route. Publish.

  • 01

    Monitor

    Subscribe to every source your newsroom cares about. Live broadcasts via RTMP/SRT/HLS, digital video feeds, text wires, social timelines, RSS, custom publications. Sources fan-in to one stream; dedup + cluster by topic in real time.

  • 02

    Score

    Each story gets a relevance and priority score across several signals. Confidence intervals attached. Customer-tuned thresholds. Scores ship as structured JSON; override any dimension via API.

  • 03

    Compose

    Drafts the article copy, social captions, video script, and headlines — tuned to your editorial voice via a customer-fine-tuned model. Every output in your 10+ supported languages. Voice model ingests your back catalog during onboarding.

  • 04

    Route

    Three triggers run in parallel: topic category, sentiment, source confidence tier. Auto-publish or editorial review — customer picks per story. Editorial workflow integrates with your existing CMS or sends to a hosted dashboard.

05

Publish

Platform-native cuts auto-generated per destination. Aspect ratio (16:9, 1:1, 9:16), length (30s/90s/3min), language, captions, brand graphics. Direct push to YouTube, TikTok, Reels, X, FB, LinkedIn, your apps, or any RTMP.

WHAT MAKES IT POWERFUL

More than a clip tool. A news desk that runs itself.

Watches the whole market, 24/7

Broadcast, video, wires, social, RSS, and custom feeds fan into one stream, deduped and clustered by topic in real time.

Surfaces by signal, not bias

Stories are scored and ranked by fit, locality, sentiment, and priority, so what is rising rises, no matter which feed an editor happens to follow.

Does what Amagi does, and more

Repurposes your own newscasts into cuts like a broadcast platform, and also watches the market and alerts you to trends before you have covered them, at a fraction of the cost.

One story, every platform

A single story becomes platform-native cuts for YouTube, TikTok, Reels, X, LinkedIn, and your own apps, with the right aspect, length, captions, and brand graphics.

Routing you control

Routine versus editorial routes on your rules (topic, sentiment, source tier), overridable any time, with a full audit trail and one-call retraction.

On the FastPix video stack

Encoding, multi-CDN delivery, player, In-Video AI, and video data underneath, so the cuts you publish are also measured and improved.

SOURCES AND DESTINATIONS

One Agent. Every input. Every channel.

What's connected on each side of the Agent — and how a single story turns into every cut you'd want to publish.

InputsMonitor every source
Live broadcast (RTMP / SRT / HLS)YouTube liveTikTok / Reels / ShortsPodcasts (audio + video)Google News / NewsAPIAP / Reuters wiresTwitter / XRedditLinkedInRSS feedsPR wiresCustom publications
OutputsPlatform-native cuts
YouTube (16:9)YouTube Shorts (9:16)TikTok (9:16)Instagram Reels (9:16)Facebook (1:1 / 16:9)X (16:9 / 9:16)LinkedIn (1:1 / 16:9)Your apps / webAny RTMP destination

Languages — auto-translated cuts and captions

EnglishSpanishPortugueseFrenchGermanItalianJapaneseKoreanMandarinHindiArabicIndonesian+ more on request

INTEGRATIONS

Plugs into your stack. Both editorial and broadcast.

The Agent doesn't replace your CMS or your production switcher. It pushes finished cuts into the tools your team already uses.

CMS + editorial integrations

For digital-first newsrooms. Auto-publish or queue items in the CMS your editors already use.

WordPressSanityContentfulStrapiCustom REST + webhook

Editorial queue can also be a hosted FastPix dashboard for teams without an existing CMS.

Broadcast + production tools

For small broadcasters. Push completed live-cut destinations into your existing production stack.

RTMP / SRT / HLS pushvMixOBSWirecastFastPix Cloud Playout

Custom production integrations (NDI, Switcher Studio, etc.) added by request during onboarding.

THE DECISION

Auto-publish or editorial review. Per story, both first-class.

The Agent doesn't pick one workflow and force it. Both are equal options. Customers configure routing per topic, sentiment threshold, or source tier — and override at any time.

Workflow A

Auto-publish

For routine, well-scored stories where speed matters and editorial judgment isn't required.

  • Story score above your thresholds across all signals
  • Source confidence Tier-1 (AP, Reuters, verified broadcast feeds)
  • Sentiment neutral or positive
  • Topic category outside the editorial-only list
  • Publishes to selected platforms within seconds; full audit trail logged
Workflow B

Editorial review

For polarizing, crisis, low-confidence, or off-policy stories where a human picks the angle and approves the cut.

  • Negative or polarizing sentiment
  • Crisis or geopolitical topic category
  • Tier-2 or Tier-3 source confidence (social, blogs, unverified)
  • Pre-cut platform-native versions ready in the editor's queue
  • One-click publish, swap, or kill from the queue
  • Already-published cuts can be retracted with a single API call; takedown trail is logged

IN PRODUCTION

Two customers running the Agent today.

Aadhan named with permission. The second customer is confidential at launch. Both have been in production with Breaking News Agent for three months.

Aadhan · Short-form news

Aadhan ships breaking news to a large multilingual audience.

"We were a tight editorial team competing against newsrooms with thirty. The Agent drafts the article and the social cuts in our voice while we are still discussing the angle."

200+
Stories/day scored
Multi-platform output
~6 hrs
Editor hours saved/day
Regional broadcaster

YouTube clips published 8 minutes after they air on TV.

"Our previous workflow needed a producer per platform. With the Agent, one editor approves the queue and every social cut is already platform-native, in our voice — the cuts are out before the broadcast segment ends."

8 min
Broadcast-to-YouTube
5
Destinations per story
3
Producers redeployed
Read the Aadhan case study →

BUILT ON FASTPIX

The Agent runs on the same video infrastructure powering your viewers.

You're not stitching the Agent to a separate video stack. It runs natively on FastPix's encoding, delivery, player, AI, and data surfaces.

Fast Encoding

Per-title encoding across Standard, Pro, Premium. Every platform-native cut is encoded once, delivered everywhere. VOD

Multi-CDN delivery

Per-session routing across 5 partners. Under-1.5s failover means breaking-news cuts stay up under traffic spikes.

FastPix Player

Open-source player embedded across your apps, web, and partner destinations. CMCD analytics baked in. Player

In-Video AI

Auto-captions, auto-tagging, voice translation, scene detection, and the customer-tuned voice model powering Compose. In-Video AI

Video Data

56-dim per-session QoE. Track which platform-native cuts drive viewer behavior, then feed back into story-scoring tuning. Data

Cloud Playout

For broadcasters: 24/7 linear channels feed the Agent as one of its source types. Playout

BREAKING NEWS AGENT FAQ

Breaking News Agent — questions, answered.

  • How does the Agent decide between auto-publishing and editorial review?

    Three routing triggers run in parallel: topic category (crisis and geopolitical topics route to editorial), sentiment (negative or polarizing sentiment routes to editorial), and source confidence tier (low-trust sources route to editorial). When all three clear customer-set thresholds, the story can auto-publish. Each customer configures whether auto is enabled at all, per source and per topic.
  • How does the Agent draft content in our editorial voice?

    During onboarding the FastPix team fine-tunes a voice model on your back catalog of articles, captions, scripts, and headlines. The Compose step then drafts article copy, platform-native captions, video voiceover scripts, and headlines in that voice for every story. The model continues to learn from editor edits made during the editorial-review path. Voice tuning is per-account; nothing is shared across customers.
  • What sources does the Agent monitor?

    Live broadcast (RTMP, SRT, or HLS pipes), digital video (YouTube live, TikTok, Reels, podcasts), text aggregators (Google News, NewsAPI, AP, Reuters wires), social timelines (Twitter/X, Reddit, LinkedIn), RSS feeds, press release wires (BusinessWire, PR Newswire), and any custom publication feed a customer plugs in.
  • How is this different from Amagi Newspulse?

    Amagi Newspulse repurposes a broadcaster's own newscasts and VOD into social clips. Breaking News Agent does that too, then adds the part Amagi cannot: it monitors the market and other participants, alerts your team to what is trending in your audience's topic area before you have covered it, and then helps you execute the clip, the article, and the publish. It also watches every source category — broadcast, digital video, text wires, social, RSS — drafts in your fine-tuned editorial voice, and is API-first at a fraction of the cost of a broadcast-grade platform. Built for the smaller newsroom that needs to punch big without a production department.
  • How does the Agent score and prioritise stories?

    Every story the Agent surfaces gets a score across several signals: how well it fits your editorial voice, its locality and relevance to your audience, its sentiment, a suggested angle, and how headline-worthy or breaking it is. The signals are delivered as confidence-scored numbers via the story.scored webhook, and the score drives both the routing decision and how each platform-native cut is composed.