VidiVidi: a table you sit down at, not a feed you fall into

In September 2024, Google shipped the feature that changed what people expect synthetic audio to sound like. NotebookLM’s Audio Overviews turned any pile of documents into a two-host podcast — warm, fluent, joking, interrupting each other — and for a few weeks the internet did little else with AI. The genre it spawned is now everywhere: personal briefings, auto-generated shows, a podcast about anything, for an audience of one. We understand the appeal — we build voice technology ourselves. But the most interesting thing about that genre is what it quietly gives up: if every listener gets their own show, no two people ever hear the same thing, and news stops being something you can talk about, because there is no “about” left in common.

What the personal feed dissolved

Broadcast media had a property nobody put on the feature sheet until it was gone: synchrony. The evening bulletin aired at the same hour for everyone, and whatever its flaws — they were real — it produced a shared object. You could disagree with it precisely because you had all heard it. Two decades of algorithmic feeds took that apart item by item, until your timeline and your neighbour’s timeline ceased to overlap; the Reuters Institute’s surveys chart the wreckage, with around four in ten people actively avoiding news and trust stuck near 40%. The personalised AI podcast, for all its craft, is that fragmentation’s logical endpoint: an audience of one, a commons of zero.

One table, one clock

VidiVidi is built on the opposite premise: a shared broadcast network, in Turkish, where the day’s agenda is discussed at one table, on one clock. Four fixed shows anchor the day — morning at 08:00, noon at 12:00, evening at 19:00, night at 23:00, all on Türkiye time — and when something big breaks, a special session called DÖNÜM convenes. There is exactly one broadcast and it runs on one shared clock: everyone who tunes in hears the same sentence at the same moment. Come late and you join at the live point, like walking into a room where the conversation is already going; replay from the start if you like — but the screen tells you, plainly, that you have left the common point. The broadcast is not generated for you and not cut for you. That is the feature.

A cast that is synthetic and says so

The table has a permanent cast: İpek anchors, Kaan reads political economy, Çetin covers foreign policy and security, Elif comes from the newsdesk, and colleagues join from their own beats. Each character has a stable role, a consistent way of thinking, a rhythm — and their own voice signature. All of it is synthetic, none of it imitates a real person, and the product says so on every screen, at the door and on every character card — the disclosure norm we argued for in our post on the voice-disclosure era, and the one Article 50 of the EU’s AI Act has made law in Europe since August 2025. The honesty extends past the labels into four commitments the product is written around: synthetic is called synthetic; a prepared broadcast is never sold as “live” — the honest phrase is “on air now”; every number on screen is a real measurement, with no fake listener counts and no fake questions; and a retracted broadcast is retracted everywhere — audio, transcript, clips and share links.

Where VidiVidi stands

Listening is free and needs no account. Membership adds one precisely scoped thing: the right to leave a question on the table. Questions are reviewed, may be merged with similar ones and — if selected — are taken up on air; a seat at the table, not an answer guarantee. Reporting an error, a legal issue or a safety concern is always free and needs no membership. Every broadcast ships with a time-coded transcript, so the spoken word stays checkable, quotable and searchable. VidiVidi is in beta — the schedule and the cast run daily, while paid membership stays closed until the payment rails open — and underneath it is the platform this site describes everywhere else: our own models and voices, through our own gateway.

• One shared broadcast on one clock — join late and you join at the common point
• Persistent synthetic cast with stable roles and voice signatures, always labelled as synthetic
• Four fixed daily shows plus DÖNÜM, the breaking-news session
• Members leave questions; the table selects and answers them on air
• Time-coded transcripts, honest metrics, retractions that close everywhere

What we will not claim

Three honest limits. First, a synthetic cast is not a newsroom, and VidiVidi does not pretend to replace one: the table discusses an agenda that originates in reporting done by real journalists at real outlets, and the product’s value is the discussion and its discipline, not a claim to original newsgathering — the evidence layer under the day’s events is Harbi’s job, and we built it separately on purpose. Second, shared-by-design means not-for-you-specifically, and we state the trade rather than hiding it: no personalised cut, no “more like this”, no show about your niche. If a briefing generated for one listener is what you need, that is a different product — our Blab voice stack is how you would build it. Third, it is a beta and behaves like one: features are missing, plans can change, and the same public guide that makes the honesty promises documents every change — including the broadcasts we retract.

The feed gave everyone their own truth at their own pace. A table gives everyone the same sentence at the same moment — and says, on every screen, that the cast is synthetic.

For listeners in Türkiye, VidiVidi is worth watching as a design position as much as a product. The loud future of AI media is personalised abundance; here is a deliberate bet that some of what news needs — a common object, a shared clock, an honest label — points the other way. The synthesis is ours end to end: the voices, the models and the gateway they run through are the same in-house stack behind Blab and the rest of this platform, which is what makes a commitment like “synthetic is labelled synthetic” enforceable in the architecture rather than promised in a policy. It is in beta and listening is free — vidividi.app, built in Ankara by Arpanet Bilişim A.Ş. Pull up a chair.