Deepening M&E Reliance on AI Fueled by More Control Over Usage

DIY Collaboration Is Transforming Operations Industry Wide

By Fred Dawson

Amid mounting global anxiety over unfettered AI development, the technology is operating in a newfound M&E comfort zone where innovative independence is generating unprecedented improvements in efficiency and monetization that seemed out of reach two years ago.

Things are happening so fast, it’s hard to assess the overall impact, but the wave of new AI-enhanced solutions flooding the exhibit halls at the recent IBC show in Amsterdam made clear the industry is leaving the old ways of doing things in the dust. “AI is something we’ve never seen before in terms of the pace of development and adoption,” said Fernando Amendola, managing director of research strategy at the M&E consultancy The Media League.

Amendola said the head-spinning outpouring of AI-enhanced products at the NAB convention in April led him and his colleagues to launch a new “intelligence-as-a-service” initiative they’re calling AI2 Insights. Acknowledging that after five days at NAB, dozens of meetings, hosting an AI forum, “I still couldn’t figure out what was happening,” Amendola said the situation was not where a consultancy like The Media League wanted to be.

“I could not give a sourced answer to people’s questions,” he noted. “And I should have been able to.” At IBC, the group rolled out AI2 as a service that can deliver answers by helping content producers and distributors evaluate AI solutions in a systematized way. Reports running to 200 pages or more deliver numerically scored analysis across multiple aspects to AI usage along with confidence ratings that register the reliability of research results generated with each project.

 MCP-Enabled AI Collaboration

 This was one of the more noteworthy one-off developments at IBC, where AI was again ubiquitously present in vendor solutions of every stripe. The over-arching difference at this event was the extent to which the industry is collaborating on freeing vendors and customer alike from over dependency on the Big Tech AI model builders.

This reflects a sudden surge in the use of open APIs with AI models and agents as embodied in the Model Context Protocol (MCP) specifications developed by Anthropic two years ago. Leveraging MCP, M&E vendors and their customers are able to custom build AI inference tools to carry out user-specific tasks using large-language and video-language models (LLMs and VLMs) without being shoehorned into pre-built products. Critically, along with enabling customization, the DIY approach avoids opening broadcasters’ databases to third-party suppliers.

Also notable as another aspect to the transformative impact of AI is how it appears to be accelerating broadcasters’ transition to IP-based production workflows, especially when it comes to sports, news and other live programming. Even though IP is widely understood to offer major advantages based on the versatility of software running on commodity processors, up to now the prevailing trend has been to pursue hybrid strategies involving shifts to IP over SMPTE 2110 in greenfield and SDI replacement scenarios while sticking with SDI until equipment wears out.

But as Imagine Communications CEO Steve Reynolds noted in a post-IBC blog, hardware-based production can inhibit producers’ ability to exploit AI. “If a facility is still moving video and audio as SDI signals rather than IP, it may be more difficult to access the AI-enabled capabilities that generated so much excitement at IBC this year,” Reynolds said. “The AI operates on data it can see and process. Signals in their baseband form are not that.”

Imagine, now doing business as a recently acquired independent operating unit of Lumine Group, has adjusted its strategy to ensure customers can use its core products while utilizing the liberating force of commonly supported MCP interfaces to inculcate workflows with AI capabilities, said Glodina Lostanlen, chief revenue officer at Imagine. By providing their AI systems secure, governed access to Imagine’s platforms through standardized gateways, “we’re able to show customers how to personalize their own experiences instead of us trying to cover everything,” Lostanlen said. “We’re going to focus on what we’re good at and make it easier for our channel partners and customer to take it to the next level,” she said.

MXL Vendor Interoperability Support Parallels MCP Impact on AI

The MCP-enabled shift to such interoperable flexibility with AI parallels another big trend that’s driving the industry in the direction of IP-based operations. This has to do with surging support for cross-vendor interoperability following the prescriptions of the European Broadcast Union’s Joint Task Force on Digital Media Facilities (DMF) and its Media Exchange Layer (MXL) specifications for real-time exchange of media across workflows.

These developments are the primary contributors, along with industry support for the next-generation MOQ streaming platform, to a historic transformation in operations where producers can seize new opportunities to serve consumers across the fragmented device landscape without having to endure the learning curves and forklifting that have long plagued service advancement.

As Grass Valley chief revenue officer Tim Banks put it at a customer gathering in Amsterdam, “Transformation has become business as usual.” DMF, a work in progress that’s formalizing standards for software-based media processing, and MXL, issued as an SDK by the Linux Foundation-sponsored MXL Project, are game changers, even for a company like Grass Valley whose Agile Media Processing Platform (AMPP) has long supported vendor interoperability across different infrastructure environments on premises and in the cloud through its reliance on

“We don’t believe customers have to make economic bets on which technologies are best,” said Grass Valley president Jon Wilson. Answering the question of “How do I modernize without disrupting what I have?” won’t be about one technology replacing another,” he added. “It’s about having the right technology in the right place as needs evolve.”

AI Everywhere Across the M&E Supplier Ecosystem

When it comes to AI, vendors across the industry are getting out the message that they’re on board with the new MCP-enabled flexibility with AI. “Operators are learning how to use AI to generate their own code,” said Ali Hodjat, senior director of marketing at Telestream.

“We’ve augmented products like our Vantage media processing platform with lots of AI,” he added, citing purpose-built AI services for use in media workflows that automate content analysis, metadata generation, quality control, content reframing and localization. “But we also recognize operators want to be more in control of how AI is used. So we’re giving them the option to license AI applications from us or use their own AI models as part of the Vantage workflow.”

But, as Hodjat noted, there’s another aspect to the DIY AI trend “toward more agentic workflows,” meaning that, as customers develop new AI agents, MCP APIs will need to be ubiquitously deployed to enable interactions with existing AI models and agents. “AI tools need to be able to interact natively on impact with other AI assets,” he said. “That’s not a reality yet.”

Meanwhile, the floodgates are open as AI innovation across the supplier ecosystem impacts content output end to end. For example, Amazon Web Services (AWS) is putting ever more effort into streamlining M&E workflows with the aid of AI.

A case in point is the AWS Elemental Inference service. The innovation is generating cost savings of up to 34% in the use of cloud processing to convert big 16x9 horizontal smart TV displays into vertical displays suitable for consumption on mobile devices, according to Stephanie Lone, global director of solutions architecture for M&E, games and sports at AWS.

That’s just part of what Elemental Inference does, Lone added, noting its support for generating contextual metadata for advertising in live video by looking at “every frame to figure out what’s in the video.” This includes sentiment analysis, which with the other metadata, is classified into key works based on the Google Ad Manager (GAM) and Interactive Advertising Bureau (IAB) taxonomies. “Our AI agents are all talking to each other to find the right ad,” she said.

Elsewhere, the many ways cloud-based post-production playout vendor Amagi is putting AI to use includes applications that represent a new positioning of the company. The new strategy comes in response to trends, especially in news, sports and other live productions, that are using the containerized microservice support provided through MXL to blend production and playout to the point of minimizing or even eliminating the traditional middleman playout role.

“We don’t perceive a threat,” said Amagi CTO and co-founder Sainidhya Srinivasan. “We see growth.”

By using MXL to launch its own live production platform in the cloud, the company is working with the likes of NBC and Sinclair Broadcast Group to alleviate the burdens of moving live broadcast production workflows to the cloud, where AI can work its magic to much greater effect. In the case of Sinclair’s live productions across an 82-market TV station footprint in the U.S., the broadcaster is using two new Amagi agentic AI services, News Pulse and Fan Pulse, which, respectively, fetch ancillary content for real-time insertion into news and sports productions.

“They specify what kinds of clips, stats, titles and other identifiers they want to insert into their productions, and our agentic AI delivers what they’re looking for,” Srinivasan said. He noted that one customer is using Amagi’s AI agents to support “smart clips,” which personalize the insertions based on profiles built from consumer data.

“We’re just beginning with our first few customers,” Srinivasan said, noting that all AI agent selections go through human review. “Currently we’re running at 70-80% approval,” he said. “Our goal is to get to 90+%.”

Flexible use of AI in production workflows is a fundamental tenant of the approach cloud-native streaming platform provider 24i is taking to use of the technology. “We understand a strong desire to maintain control over AI is shared by many of our customers,” said 24i CEO Sebastian Braun. “When a broadcaster has made a significant investment in developing their own AI solutions, they can override what we’ve built for our backend.”

But that said, Braun made clear 24i’s own AI development has gone into hyper drive. “On the whole, it’s about speed to market,” Braun said. “We’re leveraging AI with monthly releases that allow us to constantly push value to our customers.”

He also noted that 24i’s speed to market has been greatly abetted by the elimination of impediments to strategic development imposed by the quarterly numbers game played by publicly held companies. In March 24i and video device maker Amino Communications were acquired from publicly held Aferian plc by Sapphire Technology Group Ltd., which is wholly owned by U.K.-based entrepreneur Keith Todd.

Freed of vying for budgets and project prioritization at Aferian, which changed its name from Amino following its 2019 acquisition of 24i, Braun and his team have been generating new AI-enhanced modules supporting personalization, advanced advertising and much else at a rapid clip, all of which can be mixed and matched using open APIs with other vendor solutions on the 241 Video Cloud workflow. For example, he noted, 24i personalization delivers real-time, AI-driven experiences across recommendations, search, metadata enrichment, UX, and marketing.

“One of our latest releases began last year with the first version of our advanced advertising module,” Braun added. This introduced a unified server-side, client-side and server-guided ad insertion (SSAI, CSAI, and SGAI) software stack powered by data to deliver smarter monetization and better viewing experiences, he explained.

“Our first version allows customers to leverage viewership data to increase fill rates with improved performance from Digital Programmatic Network Managers (DPMNs),” he said. “We give customers the opportunity to track everything from how the campaign is set up to users’ viewing behavior, including what they watch, how long they engage, and how much revenue they generate.”

The company’s latest addition to the advertising workflow involves policy management. “We track what type of ad was sold for the various types of devices, locations of targeted users’ locations, and how the targeting is segmented by content or audience,” Braun said.

On another front, Sony is putting DMF-aligned products with MXL and AI support to work in a variety of ways that expand multi-vendor interoperability and functionalities in software-defined live productions large and small. In stride with the trend toward streamlining live production, Sony’s aim is to tie camera capture directly into production workflows across premises and public cloud locations.

“IP is the fundamental backbone connecting all these tiers of production together,” said Rob Thorne, category head for networked live solutions at Sony Europe. “We’re automating and configuring the whole production environment with global management solutions.”

Most recently, at IBC Sony’s Nevion division launched the first commercial applications of its MOXELA broadcast platform. One bridges contribution transport over unmanaged networks between on-prem and public cloud processing, and the other supports software-defined broadcasting workflows for more agile, scalable and infrastructure-independent live production environments.

All of this integrates with Nevion’s VideoIPath media orchestration platform, Thorne said. Supported functions include multiple A/V compression formats, logo insertion and audio matrix routing, he added, noting, “You don’t have to be an AWS cloud expert to support automation.”

But with all the upsides that come with the growing trend toward adapting AI LLMs and VLMs to purpose-built applications, there are big challenges related to getting smarter about using the technology and lowering the risks of misfires. As Media League co-founder Thierry Fautier put it, “The distance between what gets announced and what actually works is where budgets go to die.”

He and Fernando Amendola described the Media League’s new AI2 Insight service as a tool that could go a long way toward helping developers meet those challenges. They said confidence-rated conclusions about client projects backed by expert review are delivered to customers based on a continuous research process that currently looks from both the technology and business angles at more than 70 distinct AI techniques involving 200+ M&E use cases as monitored across more than 1,000 technology vendors and 1,000 media operators worldwide.

“Only AI can keep pace with AI,” Fautier said. “That is what makes this platform possible.”