Built around your sources, your editorial standards and your publishing channels. Genix automates the route from incoming material to published articles, podcasts and video — while your editors stay in control of every decision that matters.
A 30-minute qualifying call. No slides. We tell you honestly whether this fits.
Source
Rewrite
Fact check
Headline
Cover image
Editor's approval
Publish to CMS and Telegram
Genix grew inside a working newsroom, not a lab. Our first large deployment runs dozens of titles on one platform.
ThebottleneckisnotalackofAI
Most newsrooms already use AI. What they do not have is a process. The story sits in one tool, the draft in another, approvals in chat, the files anywhere — and the reasoning behind every change lives in one person's head.
Volume outgrew the team.
More stories, more formats, more channels, same people.
The same substance has to be reshaped for a site, a messenger, a social feed and sometimes an audio bulletin.
Production runs on individuals, not on a system.
When they leave, the process leaves with them.
Management cannot see where a piece actually stalls, or what a piece actually costs.
Search is no longer the safety net it was: publishers expect to lose 43% of search traffic by 2029 (Reuters Institute, Trends and Predictions 2026). Producing less is not an option. Producing without control is not one either.
Technologyplusprocessexpertise,notalicence
We do not rent you an empty builder and leave. We audit your content process, design the routes, configure them, connect your sources and channels, and stay while the process changes.
We work on the process first.
An empty pipeline builder produces nothing. Value appears when the route is configured for your desks, your sources and your standard.
The platform is how we do the work.
Every route you use is running software with real integrations, not a slide.
We stay after go-live.
New formats need new routes. That is the engagement, not an upsell.
Editorial teams get a production line. Management gets visibility and cost control. Legal gets a per-item record of the source, the steps, the prompts, who started the run and who published it.
Fromstorytopublishedpiece,infourstages
Bring a story in.
Start from your own material — an article URL, a video link, a file, a broadcast transcript, plain text — or, where a source feed is connected, from the News Day board, which collects incoming news into one editorial panorama with AI-scored relevance. The source feed is an integration configured during rollout, not something that ships in the box.
Run a pipeline.
A pipeline is a route you configured: which stations, in what order, where the system acts on its own and where it waits for a person.
Review at the stops.
Where the route says so, production pauses — fields to edit, an approval to give, an input to supply. Nothing moves past a stop until a person moves it.
Publish to the channel.
The finished piece goes to your CMS, your messenger channel, your social platforms, an audio catalogue, cloud storage or a document. Connectors exist for several CMSs today; connecting yours is part of the rollout, not a promise made by this page.
Every stage produces a visible result you can inspect, correct or re-run on its own — without restarting the whole route.
Thesystemstopswhereapersonhastodecide
This is architecture, not a disclaimer. Stops are configured inside the route, and you configure them — not us.
Approval gates.
Production waits for an explicit confirmation before it continues.
Manual field edits.
An editor rewrites the headline, the lead or any field mid-route, and the route continues from there.
Input stops.
The route waits for your text, your files or your list of links, and transcribes attached audio if needed.
You set the automation level.
Where the risk is higher, keep more manual decisions. Where the scenario is proven, remove them. That call belongs to the newsroom, not to the vendor.
43%accept news made by a person with AI help, only 12% accept news made mostly by AIReuters Institute, Generative AI and News 2025, six countries
Builtinsideanewsroomthatshipseveryday
Genix did not start as a product looking for a newsroom. It started inside a media group that had to publish across dozens of titles with the same headcount, and it has been part of that group's daily production ever since.
41titles run on one platform, each with its own routes, prompts and limits
94production routes in active use
56step types to assemble routes from
11,000+production runs a month
11,561items published through the platform
4,154press conferences and videos transcribed a month
3,082channel adaptations a month
Figures are a snapshot of our first large deployment, taken 26 August 2026. They describe one media group, not an aggregate customer base, and they are not a promise of the same result for you.
≈10%autonomous agentic use of AI is still rare in newsrooms, even in the most mature use caseHere it is daily production.WAN-IFRA · FT Strategies, 2026
Whatmakesthishardtocopy
Traceability of every piece.
Source → steps → prompts → editor's approval, recorded per item. Not a policy on paper — a record in the system. A tool whose unit of work is "one generated text" cannot show a chain its data model never had.
Live broadcast to published piece.
The system listens to live air, transcribes it, detects where one story ends and the next begins, and starts the route on its own. Everyone transcribes; the automatic path from speech to publication is the part we have not seen in open practice.
One platform, many media units.
A new title, a new vertical or a new language version is configuration, not a development project. Each unit keeps its own routes, prompts, provider keys and spending limits. Isolation today is at the level of a media unit inside one deployment — we say that plainly rather than calling it "without limits".
A fact-check step inside the route.
The AI output is checked against the source material and unsupported claims are flagged. Whether that step blocks publication or only flags issues is a per-route setting the newsroom controls, and the run page shows which setting was active. Elsewhere fact-checking is a separate tool someone has to remember to open.
We record what the system did, from which source, with which prompts, who started the run and who published the result. There is no per-step audit trail of exactly who approved which gate and when. Do not build a legal argument on this line.
Seewhereyournewsroomsitsagainst227usecases
We compared our own platform against 227 types of editorial AI use documented worldwide. Then we turned the method into a product. You get a map: what already works, what you can switch on with a prompt, what needs integration, and what is not worth the effort.
Several titles under one roof, each with its own standard, tags and covers.
High-tempo and broadcast newsrooms.
Hourly bulletins, live air, several output formats from the same material.
Brand media, publishers and content agencies.
Several voices, several clients, unit economics that has to survive scale.
If you publish occasionally, work alone, or want AI to write without an editor, we are the wrong vendor. We say so on the call rather than after the contract.
Wherewestop,onpurpose
Genix ends at publication. Reader personalisation, reader-facing chatbots, comment moderation, paywalls and monetisation, content litigation and deepfake detection are outside the product. That is a decision, not a backlog.
Genix does
Genix does not promise
Moves material through a managed production route
Does not replace your editorial policy or define quality without a person
Makes stages, results and history visible per item
Does not remove fact-checking, editing or brand responsibility
Helps you scale output and formats
Does not guarantee an outcome by containing AI
Runs alongside your existing systems
Does not require you to replace your current stack
Whatyourlegalandsecuritypeoplewillask
Human decision points are architectural. Approval gates and manual stops are steps in the route, configured by the newsroom and visible on the run page — not a vendor policy that changes without you.
Every piece has a history. Source, steps, prompts and approval, recorded per item.
Provider keys are encrypted (AES-256-GCM) and access to a specific key can be restricted to specific media units.
Roles are scoped per media unit. Permitted routes, AI profiles and prompts are configured per unit too.
AI spending has limits set globally and per unit, enforced by the system rather than discovered in an invoice.
Start with the process, not with a licence
The first call is 30 minutes and free. We look at what you publish, how much of it, and where it stalls — and tell you whether a content process audit is worth your money. If it is not, we say so.