Broadcast Media: The Inside Track

🎙️ Earlier this month, one developer used AI coding agents to clone Photoshop, then Lightroom, then Word, Excel and AutoCAD, in a matter of weeks. Another built a Photoshop-style editor for around two thousand dollars in AI usage. In the same month, Anthropic filed for what could be the biggest IPO in history.

Ben and RAIANA ask what happens to an industry when building software becomes cheap, and who actually captures the value.

🔍 In this episode:

• The vibe-coded Adobe clones: impressive demo or real revolution, and why the last ten percent of polish is where the years go
• Sam Altman’s betting pool on the first one-person billion-dollar company, and why it sounds less like a joke in 2026
• From Meta, Uber and Airbnb to OpenAI and Anthropic: network unicorns versus companies selling intelligence by the unit
• Anthropic’s IPO by the numbers: a reported raise of up to $100bn, valuations of $1.5–2tn, revenue of $16.3bn in H1, and an operating profit in Q2
• The other side of the ledger: more than $500bn committed to chips, data centres and energy, and what public shareholders will expect in return
• Value concentrating at the top while the cost of building falls toward zero, and why the middle gets squeezed
• Three moves for broadcasters: use your vendor renewals as leverage, replace the small internal tools first, and treat governance as non-negotiable
• Where the people go: why knowing what to ask for, and knowing when the answer is wrong, is the new craft

🧠 The thread

For thirty years building was expensive and ideas were cheap. Now building is cheap, and good judgement and trust are the scarce things. The question for every broadcaster isn’t whether you can afford to build any more. It’s whether you know what’s worth building.

📍 Next time

Ben reports back from AI Expo Africa at the Sandton Convention Centre in Johannesburg.

🔗 More info on Ancast:

www.ancast.co.uk

🎧 Listen & subscribe!

#BroadcastMedia #AI #VibeCoding #Anthropic #OpenAI #IPO #SoftwareDevelopment #MediaTech #AIExpoAfrica #InsideTrack

🎙️ Fresh from IBC, Ben lands at his first ever CABSAT in Dubai, at its new home in Expo City and its new October slot. In this episode, Ben and RAIANA unpack what the week taught him: a mature streaming market, sovereign AI as the talk of the opening day, and the exam question every newcomer asks: can I really do business here in broadcast AI?

🌍 What we cover

• Why the MEASA streaming market is setting the pace, not catching up
• Sovereign AI on the opening day: the honest admission that the OTT platforms are way ahead of traditional broadcast on machine learning
• What streaming economics sounded like: retention, discipline and not buying your growth
• Ancast Nowcasting™: real time audience prediction, and why it needed a custom model
• Dubai beyond the halls: a thriving OTTRED coworking day, open book contacts and liquidity
• Easy to set up, harder to win: the exam question for newcomers
• The scorecard: do conversations become pilots, and pilots paid work?
• 
  

🛠️ Three takeaways

• Go to learn first
• Treat sovereign AI as a live buying criterion, not a slide
• Don’t confuse easy setup with easy wins
• 
  

📩 In scheduling, FAST or broadcast AI in the region? Ben’s DMs are open.

🔗 Listen on your platform

#CABSAT2026 #CABSAT #SovereignAI #Broadcast #MediaTech #OTT #Streaming #FAST #AIinBroadcast #Nowcasting #Dubai #MENA #ContentScheduling #Ancast #InsideTrack

🎙️ At IBC this year, a company called CAMB.AI demonstrated something that used to take hours: a live football commentary feed, dubbed, lip-synced and voice-cloned into another language, in seconds.

Ben and Raiana dig into what's actually changing when localisation gets cheap and fast — and it isn't what's on screen.

🔍 In this episode:

• CAMB.AI's live dubbing tech at IBC, already deployed by Ligue 1+, NASCAR and Eurovision Sport, and why sport got there first
• Why the real story isn't the demo, it's the contract — minimum guarantees starting to separate localisation cost from rights value
• A hard look at the "ten to fifteen per cent discount" number everyone's repeating, and why the evidence for it doesn't actually exist yet
• Amazon Prime Video's twenty five new theatrical localisation languages, and what that tells you about where platform leverage is heading
• The voice-cloning consent question nobody's negotiating carefully enough
• Three things a rights holder should check in their contracts this week

🧠 The thread

Good dubbing was always meant to be invisible. This year it got cheap enough to be everywhere — and the part of the industry worst at tracking change is the part that only watches the screen.

🔗 More info on Ancast: www.ancast.co.uk

#BroadcastMedia #AIDubbing #Localization #IBC2026 #SportsRights #StreamingEconomics #MediaTech #InsideTrack

🎙️ Ben has just got back from IBC2026. Forty one thousand people, a hundred and seventy countries, fourteen halls, and a trade press that agreed on the headline before the doors opened: AI has moved from demo to deployment.

That is broadly true. It is also not what Ben remembers.

What he remembers is five conversations, and almost none of them happened on a stand. This episode is a personal review of the show through those five moments, with Raiana pushing on what each one actually means for anyone who builds, buys or runs broadcast technology.

🔍 In this episode:

• The temperature of the room: vendors pitching orchestration instead of boxes, IABM re-emerging as the International Association of MediaTech, and an AI actor taking questions from a live audience at the opening keynote
• #1 The after-show drinks Ben nearly skipped, and why one evening in a room did more than twelve months of email
• #2 A CTO's questions: not "can it work" but "where does it break, who owns the data, and what happens at three in the morning when it's wrong"
• #3 Bumping into a fellow startup founder in an aisle, and why in broadcast your peers are neighbours, not competitors
• #4 Sovereign AI, raised over coffee by a senior big-tech leader and echoed on the EBU stand: ORF, ARD, Schwarz Digits, NVIDIA and Ateme all showing what "owned intelligence" looks like for public service media
• #5 Audience measurement, the least glamorous topic on the floor and the one that moved furthest, from polite interest to "when can we meet"

🧠 The thread

The interesting layer of this industry has moved. For twenty years the action at IBC was in the big halls with the big vendors. This year it was in the seams between them, and underneath every conversation was the same question: who owns the model, who owns the data, and who owns the relationship with the audience.

🎯 Three things to take back to the office if you didn't make the trip: the question to ask every AI vendor, the workflows you should list before your board asks about model dependency, and how quickly you really know what your audience is doing right now.

📍 Mentioned

IBC2026, 11–14 September, RAI Amsterdam
Particle 6 and the Tilly Norwood keynote
IAMT (formerly IABM)
EBU, ORF Innovation, ARD Cloud Unit, Schwarz Digits, NVIDIA, Ateme, Scaleway, Qvest

🔗 More info on Ancast:

www.ancast.co.uk

#IBC2026 #BroadcastMedia #AIinBroadcast #SovereignAI #AudienceMeasurement #MediaTech #PublicServiceMedia #InsideTrack

🎙️ A bonus episode recorded owith a guest panel from a small scale 'Nomad Basecamp' conference in Pokhara, Nepal - nine days before Ben lands at IBC in Amsterdam with forty five thousand other people.

Two very different guests, one question: what are these things actually for?

Ben is joined by Sheikh Saifuddin, investor, speaker and founder of Quantum Self, and Biplab Karki, founder and CEO of Hyperce, building Clicks Cloud - an agentic AI e-commerce platform out of Kathmandu.

🔍 In this bonus episode:

• Two theories of a room: forty five thousand people, badges and carpet in Amsterdam, versus a room full of people, a volunteer day and a summit circle in Pokhara — both filed under "events"
• Ben's framework put to the panel: a talk is reach, a conference is peers, a trade show is procurement — and whether booking the wrong one is why people come away disappointed
• The myth of the deal on the stand: five to nine touches before anything closes, and why the contact list was always the real return
• Indirect ROI - branding, footage, the social post - and the honest three-to-six month lag on anything direct
• Agentic commerce: why Biplab picked the boring, hard industry, and what happens when the store's agent talks to the ERP's agent
• AI as "another mind" - Sheikh on using it as an on-demand advisor to find the blocks you cannot see yourself
• Ancast Nowcasting: not an LLM wrapper, but a custom model on a broadcaster's own data plus live public signals, giving schedulers historic pattern and current intent
• What each of them is chasing over the next twelve months

🧠 The thread

Everyone in the room had a different reason for being there, and none of them was the reason printed on the ticket. The people who get value from events are the ones who worked out what kind of room they were walking into before they booked the flight.

🔗 Guests

Sheikh Saifuddin — Founder, Quantum Self · CEO, Nobl ArtLinkedIn: https://www.linkedin.com/in/sheikh-s-5440a527/Instagram: https://www.instagram.com/quantumsheikh/TikTok: https://www.tiktok.com/@quantumsheikh

Biplab Karki — Founder & CEO, HyperceLinkedIn: https://www.linkedin.com/in/beeploveInstagram: https://www.instagram.com/beeplove_/Company: https://hyperce.ioProduct — Clicks Cloud: https://clickscloud.io

📍 Upcoming

Now - Nomad Basecamp Nepal 2026 PokharaIBC2026: 11–14 September, RAI Amsterdam where Ben is taking meetings; reach out if you want one.

🔗 More info on Ancast:

www.ancast.co.uk

#BroadcastMedia #IBC2026 #NomadBasecamp #TradeShows #AgenticAI #AgenticCommerce #Nowcasting #MediaTech #Startups #InsideTrack

On the 13th August a new channel appeared in the live lineup on The Roku Channel. It has a name, a schedule and ad breaks. It looks like a television channel in every structural sense. And it is the first one in the world where nobody is in it.

Within forty eight hours the internet had settled on a phrase for it: AI slop. Ben's argument is that arguing about whether it is any good is the least interesting conversation available.

Ben and Raiana take Fairground AI Creator TV apart, and then connect it to a much bigger claim doing the rounds on the AI summit circuit: that within eighteen months we will see a company that runs itself.

🎧 In this episode

• Fairground AI Creator TV, founded by the former CEO of Xumo, four million dollars of seed funding, and why the founder detail matters
• Why quality is the wrong argument and the cost curve is the real story
• What happens to free ad supported TV when content has no acquisition cost, no rights window and no residuals
• Infinite supply against fixed evening attention, and why that is a pricing story rather than a technology one
• The part of the market actually exposed: not prestige drama, but the enormous reliable middle where most of the industry works
• The eighteen month autonomous company claim, and the numbers on both sides of it
• Task specific agent adoption heading toward forty per cent, and over forty per cent of agentic projects forecast to be cancelled

🧠 The wall nobody at the summit mentions

A model can execute the work of a business. It cannot be the business. Legal personhood needs a person, a bank account needs an owner, a contract needs a party that can be sued. So what is arriving is not a company with no humans. It is a company with one human, whose entire remaining job is accountability.

💡 Three things to do on Monday

• Be honest about which part of your output exists to fill a schedule rather than to be sought out
• Stop evaluating this on quality and start evaluating it on cost per hour
• Write down where accountability sits in every automated process you already run. Not where the technology sits. Where the signature sits

The channel with nobody in it still has somebody behind it. It always did.

🔗 More info on Ancast:
www.ancast.co.uk

#BroadcastMedia #AI #MediaTech #FAST #StreamingTV #GenerativeAI #AgenticAI #Accountability #Broadcasting #InsideTrack

There is a version of this industry that exists only on a trade show floor. Synthetic anchors, automated newsrooms, a presenter who does not exist telling you about a technology that will change everything. Then you walk into a real media operations room at three in the morning and none of it is there.

What is actually running is a transcription job, a tagging engine and a compliance flag waiting for a human to agree with it.

Ben and Raiana take an honest stocktake of AI on the content side of broadcast. Not scheduling, not the business layer. What happens to a piece of video between the moment it is shot and the moment it goes out.

🎧 In this episode

- The adoption reality: a quarter of broadcasters in, up from nine per cent, and forty eight per cent with no plans at all

- The inversion nobody talks about: metadata, subtitling and compliance shipped, AI anchors did not

- Metadata and archive search, and why the model matters far less than the state of your data

- Why you should buy on precision rather than recall, and never trust a tool that hides its confidence scores

- Compliance as triage, not judgement, and where accountability really sits

- Localisation: one transcription pass, three deliverables, and the human who stays for sarcasm

- Sport as the proof case: nine matches, one or two operators, clips in real time

🧠 The pattern behind every success

High volume, a cheap and obvious failure mode, and a human sign-off gate that already existed. The deployments that worked were not the ones with the best technology. They were the ones that landed in a workflow that already knew how to say no.

💡 Five things to do on Monday

- Choose your first use case by finding the existing sign-off gate

- Fix taxonomy and rights metadata before you buy anything

- Evaluate on your own content and your own edge cases

- Demand confidence scores in the interface

- Write down where accountability sits before the pilot, not after the incident

🔗 More info on Ancast:

www.ancast.co.uk

#BroadcastMedia #AI #MediaTech #PostProduction #Metadata #Compliance #Localisation #MediaAssetManagement #Broadcasting #InsideTrack

📺 WHAT IS BROADCAST NOW?

Ask a media studies class fifty years ago and you got a tidy list: five mass media — print, outdoor, cinema, radio and television. Broadcast was the last two. One signal, one tower, everyone watching the same thing at the same time. That list survived world wars, colour, satellite and cable. It felt as permanent as the periodic table.

Then a sixth medium turned up. We filed it under "new media" and told it to wait its turn. In 2026, it isn't waiting — it's eating the table. This episode, Ben and Raiana take on the question the industry keeps dodging: what actually IS broadcast now?

🔍 IN THIS EPISODE:

• The numbers that should worry every commissioner: fewer than half of UK 16–24s now watch ANY broadcast TV in a given week — down from 76% just eight years ago
• "Ben, you do realise that to a lot of millennials, TikTok IS broadcast" — the line a colleague said once that Ben has never been able to unhear
• YouTube is now the single biggest distributor of television in American living rooms — bigger than any network, bigger than Netflix — and it's the first stop when young Brits switch on the actual telly
• Ben's view from inside the machine room: how his projects went from racks, wires and aerials to cloud transformation and virtualised playout — and the moment he realised Silicon Valley had quietly become the transmission chain
• From PewDiePie shouting at video games to a creator earning $300 million a year: how the industry laughed at the people who became broadcasters
• The billion people a month watching podcasts on YouTube — so much for the "short attention span" theory
• Vertical micro-drama: 60-second episodes, phones held upright, and an eleven billion dollar business that offends every 16:9 instinct Ben has
• ITV, Channel 4 and the BBC uploading full shows to YouTube — rational strategy, or feeding the crocodile and hoping to be eaten last?

🤖 THE AI ANGLE:

The algorithm is the new transmitter — and what's feeding it increasingly isn't human. One in five Shorts recommended to new users is now mass-produced AI "slop," with automated channels racking up 63 billion views. Ben makes the case that in a feed full of anonymous synthetic content, a named human who stands behind the output is suddenly the scarcest asset in media — and the one thing broadcasters still own.

🎯 PLUS: Three Monday-morning moves for anyone running a channel, a studio or a newsroom — and Ben's final answer to what the word "broadcast" actually means when the towers are gone.

#BroadcastMedia #FutureOfTV #YouTube #TikTok #VerticalVideo #MicroDrama #AI #Streaming #CloudTransformation #MediaIndustry #CreatorEconomy #Podcast

🤖 In 2018 the industry laughed at Xinhua's stiff digital newsreader - creepy, gimmicky, never going to happen. Eight years on, nobody is laughing. India Today's award-winning AI anchor Sana delivers bulletins several times a day in multiple languages and has clocked up hundreds of hours of programming. Grupo Fórmula in Mexico runs a whole synthetic family - NAT for entertainment, SOFI for politics, MAX for sport - with videos clearing a million views. Chinese regional stations are broadcasting digital clones of their own star hosts. And industry watchers say this is the year we see full news programmes with no human presenter at all.

And yes - a podcast hosted by a man actually surnamed Anchor was always going to have to cover this one.

🎙️ In this episode, Ben and Raiana get into:

• The moment Ben saw this coming: robotic pedestal cameras gliding around the studio floor with no operators — the industry automating everything around the presenter until the anchor was the last human left
• The real business case: scale, speed and consistency — five or six fronted stories a day, per avatar, in every language your audience speaks, at a cost no human roster can match
• The trust problem: audiences accept AI in the plumbing but their comfort collapses the closer the machine gets to the camera — and behind Mexico's NAT there are still five humans writing, checking and producing every story
• Governance before glamour: why a synthetic presenter must be a reader, never a writer; how generative models hallucinate "with perfect newsreader confidence"; and why every sentence needs a named human sign-off
• Disclosure and the law: the EU's transparency rules land in August — label the synthetic, sign the real, and never let the audience guess. The danger isn't the labelled avatar; it's the unlabelled clone
• The apprenticeship problem: broadcast schools full of students training for the entry-level rungs the machine eats first — you don't lose the star, you lose the path that made the star
• Where synthetic presentation genuinely earns its place: hyperlocal weather, travel, markets, sports round-ups, overnight bulletins, multilingual versioning, signing avatars for accessibility
• And what it costs: the weather presenter transmitting delight, not data; gravitas on the terrible nights and simple human joy on the daft ones. The machine can describe the rain — it cannot enjoy it
• The Monday morning three: write your disclosure policy before you commission the tech, draw the transmission/witness line in writing, and get likeness rights into talent contracts today — your star's digital double is a rights negotiation waiting to happen

🧭 The anchor is not the thing that moves. It is the thing that holds you steady while everything else does.

🎧 More about Ancast:
www.ancast.co.uk

#BroadcastMedia #AI #AIAnchor #SyntheticMedia #Newsroom #BroadcastAI #MediaTech #TrustInNews #AIGovernance #ContentAuthenticity

🎙️ Image, text, and now audio have all crossed the line into the genuinely indistinguishable. So the question stops being "can I spot the fake?" and becomes "can I prove where this came from?"

Ben and Raiana get into content provenance — the quiet infrastructure that may matter more to broadcasters than any single AI model.

🔍 In this bonus episode:
• Why detection is an arms race you're built to lose, and provenance is the flip that wins
• What C2PA / Content Credentials actually does — a cryptographic chain of custody for media
• The Macron plane clip and the "liar's dividend": when real footage gets waved away as AI
• Metadata gets stripped on upload — how durable provenance and live-stream signing answer that
• Who's already shipping it: Samsung, Pixel, Canon, Sony, plus LinkedIn, TikTok and YouTube
• The EU labelling rules landing this August, and three things a newsroom should do on Monday

🎧 Listen & subscribe:

#BroadcastMedia #ContentProvenance #C2PA #ContentCredentials #Deepfakes #MediaTech #AIinBroadcast #TrustAndSafety