IBC 2026 Preview: AI is no longer a bolt-on feature but a mandate

AI is becoming an operational tool rather than an experimental feature in production.

At IBC 2026, running Sept. 11-14 in Amsterdam, coverage of AI is expected to focus on practical deployments across production, newsroom operations, post-production, asset management and localization, with attention to governance, integration and measurable outcomes.

Bolting on AI won’t cut it

“The industry is about to have an uncomfortable realization: bolting an AI feature onto a legacy platform doesn’t get you very far. AI has to be native to the product, not sitting on top of it. At Setplex, that’s the standard we apply to everything we build. Every new capability, from search to aggregation to ad automation, has AI built into its core rather than added as an afterthought. The platforms that make this shift will pull ahead; those that continue treating AI as just another feature to bolt on will fall behind,” said Lionel Dreshaj, co-founder and chief executive of Setplex.

Derek Barrilleaux, chief executive of Projective, raised a related warning — not about legacy platforms, but about disorganized ones.

“If your workflows are chaotic, AI does not fix the problem, it amplifies it. The real gains come when media intelligence runs inside a governed project framework, turning huge, untagged libraries into something editors can search, clip and drop straight into the timeline without hopping across tools or breaking security. It’s time to stop treating AI as another bolt-on point solution and integrate it as an efficiency engine that only delivers when the underlying structure, metadata and permissions are already under control,” Barrilleaux said.

Skepticism toward the generative hype

Not every executive is convinced the flashiest AI applications are the ones worth watching.

“As enthusiasm for AI drops and repetitive ads push viewers toward cancelling, the AI worth betting on isn’t the generative kind on the show floor — it’s the quieter, more practical kind that caps frequency and catches quality issues before they lead to churn,” said Karl Tempest-Mitchell, executive vice president of Setplex.

That same wariness extends to how AI’s output is actually judged, not just which applications get funded.

Advertisement

“As AI moves into production, the question nobody is really asking is whether the AI actually improved the picture. Conventional metrics were built to reward fidelity to the source, but AI does the opposite, generating detail that was never there. That’s why subjective testing matters more now: you only really know a change works when real viewers say the result looks better,” said Dani Megrelishvili, chief product officer of Beamr.

Governance becomes the harder problem

As AI use expands, several executives said the operational challenge has shifted from deployment to governance.

“AI has moved from demonstration to production faster than the industry’s ability to measure it, and media is not exempt from that. In the workflows we analyse, the cost of an AI process varies enormously depending on the model chosen and how it is invoked, and teams are routinely surprised by which part of the pipeline turns out to consume the most energy and money. Governance in 2026 should mean knowing the cost, the energy and the business outcome of every inference, not simply deciding which vendor’s model to trust,” said Kristan Bullett, chief executive of Humans Not Robots.

A parallel governance question is playing out over what AI is allowed to touch, not just what it costs to run.

“AI is table stakes now, but almost nobody has defined what it looks like to implement it with the same diligence we apply to any other production system. Hybrid AI workflows are early, and they carry governance questions sharper than the ones cloud raised a decade ago, because the issue is no longer where data sits but what is permitted to process it. The deployments we see working treat the data platform itself as the gatekeeper, using the same policy engine that already moves the data to decide what gets enriched and where, rather than bolting governance on through more tools and more steps,” said Orlando Richards, head of product at Pixitmedia.

That question of access extends to how AI agents interact with an archive once they’re turned loose on it.

“AI has moved from experiment to operational tool, but the deployments delivering real value are unglamorous — tagging media with rights and metadata, surfacing key moments in a clip, producing accurate transcripts and translations. The harder problem now is governance: when an AI agent searches an archive, permissions need to be checked at query time rather than only at index time, and its access needs to be scoped and revocable. Buyers at IBC should be asking vendors how agent permissions are inherited and enforced, because that determines whether AI can touch production media at all,” said Kathleen Barrett, chief executive of Backlight.

Ali Hodjat, senior director of marketing at Telestream, tied the governance debate back to trust more broadly.

“AI is rapidly evolving from task automation toward orchestrated, agent-driven workflows, but governance and trust are becoming just as important as the AI itself. As organizations introduce more AI into production, the differentiator will be how effectively these capabilities are orchestrated across existing workflows while maintaining operational control and transparency. Success will depend not on how much AI organizations deploy, but on how effectively they govern, integrate, and operationalize it,” Hodjat said.

Grounding the abstraction in concrete tools

Beyond the governance debate, several executives pointed to specific deployments already changing daily workflows.

“AI’s decisive move in production workflows right now is content-level validation — checking whether a frame is genuinely clean or a caption is actually accurate, using the audio and video track as the reference — rather than just confirming a file exists. On QC, systems are learning to flag which streams need full frame-by-frame inspection instead of checking every asset identically, while advanced ASR allows direct transcription against the audio track to catch drift and misheard dialogue prior to delivery. We’re building that logic into Baton, where operator corrections retrain the model so a recurring error gets fixed at the source instead of resurfacing across downstream assets,” said Anupama Anantharaman, vice president of product management at Interra Systems.

Craig Wilson, principal enterprise specialist for broadcast at Avid, made a related point about where AI needs to sit relative to existing tools.

Advertisement

“AI becomes most valuable when it’s integrated directly into the tools people already use every day, rather than sitting alongside the production workflow. It can analyse footage to enrich metadata, enable natural-language search across large archives, surface relevant material as stories develop and generate temporary visuals that help teams communicate creative intent without interrupting the edit and automate many of the repetitive tasks that slow editors down. Crucially, organisations need to retain choice over the AI models they use, while people must ultimately remain firmly in control of every creative and editorial decision,” Wilson said.

Costa Nikols, strategy advisor for media and entertainment at Telos Alliance, said today’s AI deployments are concentrated in tasks such as captioning, language detection, translation and content classification, where speed and consistency matter more than subjective judgment.

“Governance and integration remain essential: models must be auditable, predictable, and aligned with editorial standards. While AI will eventually support more adaptive production decisions, latency, determinism, and production bias still limit real-time creative automation. The near-term value lies in augmenting human operators with reliable, scalable tools that streamline workflows and enhance accessibility,” Nikols said.

AI reaching beyond the edit bay

A handful of executives pointed to AI’s growing role connecting production content to teams that never touch a timeline.

“Legal, finance, marketing, brand compliance, and rights-management teams have always needed information from media content; the tax was that every request often meant another duplicated copy handed over the wall. AI has removed that tax by working with the content where it lives, extracting enriched metadata and connecting to other business software through governed connections built on open protocols like MCP. For teams outside media operations, these connected tools can check a contract term, ensure brand compliance, feed rights reporting, or confirm contract fulfillment without a single asset leaving the library,” said Rick Capstraw, chief revenue officer at Signiant.

A similar shift toward infrastructure-level integration is underway at LucidLink, according to Samer Kamal, head of content, brand communications and product marketing.

“AI in production is shifting from standalone experiments to infrastructure-level integration, and the deployments gaining traction, like our Python SDK and MCP server, let AI agents work directly with media assets inside Filespaces rather than requiring separate extraction or duplication. That matters because broadcasters remain cautious about AI, and the tools sticking are the ones plugging into existing storage and asset pipelines, including partners like Iconik for AI-driven metadata and search, rather than asking teams to adopt a parallel system. At IBC, watch for AI features framed around governance and provenance rather than automation speed alone, since that’s what’s actually unlocking adoption inside newsrooms and post houses,” Kamal said.

Sarah Hackforth, international sales director at Big Blue Marble, described a related shift toward analyzing content once and reusing the results many times over.

“AI is becoming more valuable in production workflows as broadcasters move towards a shared approach to media intelligence, where content is analyzed once and the results are reused across several applications. For instance, a transcript created for captioning can also support chaptering and highlight creation, and visual analysis used for search can help reframe content for different aspect ratios. Bringing speech, visuals and on-screen text together also makes it easier to identify inconsistencies that may be missed when each signal is analyzed separately,” Hackforth said.

From assistant to actor

“The meaningful shift at IBC 2026 will be from AI as an isolated assistant to AI as an operational layer across all media workflows. The real test is whether systems can go beyond answering questions or offering advice and take useful action within relevant business parameters such as content rights, editorial policies, scheduling priorities, advertising requirements and distribution specifications. Buyers should be wary of point solutions that lack context beyond a single task,” said Jon Cohen, chief revenue officer at Frequency. He said the most successful deployments will be measured by outcomes such as time saved, errors prevented and revenue generated, not by how many AI features a platform lists.

Advertisement

Kristie Fung, senior vice president of product and program management at TMT Insights, framed that same shift more bluntly.

“The conversation around AI has shifted from ‘Does your platform have AI?’ to ‘Where does it deliver measurable operational value?’ Buyers should look beyond standalone AI features and focus on solutions that are embedded into production workflows, reducing manual effort, and shortening turnaround times without adding operational complexity. Localization is one of the clearest examples, where AI is helping organizations scale translation, dubbing, metadata enrichment, and quality control with far less effort, but the real differentiator is how quickly those capabilities can be deployed and consistently deliver results across large content libraries,” Fung said.

Bartell Cope, vice president of go-to-market at Lingopal, said localization is also where some of AI’s most mature deployments already live.

“AI is becoming a part of everyday production infrastructure, with AI-powered localization and accessibility workflows — such as transcription and translation — being some of the most well-established and mature deployment areas today. Investment is being driven by rights holders’ growing need to extend global reach, deepen viewer engagement and maximize ROI from content. Innovation in real-time translation is making content more globally accessible by allowing broadcasters to serve audiences in their preferred languages, without losing the original speaker’s voice, emotion or authenticity. The greatest value comes when AI integrates seamlessly into existing workflows that broadcasters already trust, while maintaining editorial quality, low latency and operational reliability,” Cope said.