Agentic AI development in Media Tech
What the latest products and services reveal about the future of the industry
Instead of adding to the flood of IBC reflections already on social, I want to use this space to explore a specific topic: Agentic AI. This is a topic that has been on my radar for a while and I believe deserves a closer, more analytical look.
In the last edition of Sense the Frame, I reflected on how AI has significantly accelerated technology development cycles, and how this acceleration is emerging as a key driver of M&A activity.
I have re-shared the post where I talk about this below, which includes my pre-IBC reflections 😊.
I believe that the recent rise of Agentic AI in Media Tech is a great illustration of this trend. While Agentic AI is still very much at the bleeding edge of our industry, the pace of progress is striking compared to historical standards. What once might have taken years is now being achieved in a matter of months, with significant advancements in technology supply emerging in less than half a year.
Agentic AI uses sophisticated reasoning and iterative planning to autonomously solve complex, multi-step problems
Source: NVIDIA Blog
This is why I have decided to focus on this topic in this edition of Sense the Frame, which includes:
A brief history of Agentic AI in Media Tech (as I’ve witnessed it).
A roundup and analysis of the latest Agentic technologies and innovations in Media Tech (and the trends they might reflect).
As usual, if you have any feedback about Sense the Frame, feel free to send me a message:
I hope you enjoyed my analysis. If you found it valuable, please consider sharing this article on your social media platforms. It helps spread the word and supports the time and effort I put into creating this newsletter. And if you haven’t already, I would love for you to subscribe so you don’t miss future editions.
Brief history of Agentic AI in Media Tech
Let’s first recap the trajectory of Agentic AI in Media Tech in 2025:
The conversation about Agentic AI in Media Tech first emerged after the keynote of Nvidia’s CEO Jensen Huang at CES 2025, where he spotlighted agents as the next big thing for technology-driven industries, alongside other breakthroughs in AI. At the time, many of us were sceptical about how much impact this technology could realistically have within the same year (due to how long the industry normally takes to absorb macro technology trends). In my 2025 industry outlook, I argued that it was perhaps too early for Agentic AI to leave its mark on the industry this year (yes, I was probably wrong 😊).
NAB Show 2025 seemed to confirm that prediction. There was little discussion of Agentic AI, and only a few products or features on display made tangible use of the technology. Yet, in the weeks following the show, I noticed a (slightly) rising wave of press coverage on the technology and its potential impact on the sector. Much of it, in my view, leaned toward hype rather than substance (something I highlighted in my summer half-year review of industry trends).
In that same review, however, I made it clear that I was open to being proven wrong by new announcements at IBC 2025. And the summer (particularly the period between July and August) brought a wave of product announcements, signalling a rapid acceleration in the development of Agentic AI in the Media Tech industry. A key driver of this was likely the introduction of the A2A protocol in April 2025, which lowered the barriers to developing Agentic AI systems. These products, many of which were announced for and showcased at IBC 2025, represent the focus of the following analysis.
The figure below is a graphical illustration of the historical progress of Agentic AI in Media Tech in 2025, based on my observations.
Roundup of Agentic AI in Media Tech
Before diving in, a few words of caution regarding the scope of this analysis:
This roundup is based on my review of product and service announcements (heavily) featuring Agentic AI between June and September 2025. I did not have the opportunity to see these technologies in action while at IBC. Also, apologies if I have missed any announcements (feel free to reach out to me with any feedback).
The analysis is primarily focused on applications of Agentic AI in broadcast and media. The main objective is to identify where the technology is being applied and, from that, draw insights on potential industry hotspots and directions (I focus on these at the end of this piece if you want to skip to them).
This is an analysis of technology supply, not demand. It reflects where technology suppliers are investing, but it does not necessarily indicate how buyers will respond.
With this context in mind, let’s start upstream in the content supply chain and work our way downstream.
Upstream: news production
Content production shows the least Agentic AI penetration. This is not surprising as the application of AI in content creation remains very controversial, with companies (technology suppliers and media businesses) generally thinking that they much more to lose than to gain in this area compared to other segments of the content supply chain.
The only relevant and sector-specific (i.e., I am not considering generic Gen AI video creation tools marketed by big tech companies in this analysis) application I could find is HighField AI. Introduced at NAB Show 2025 but officially launched in July 2025, this is an “Agentic and multimodal AI platform for automating graphics production,” with a focus on news production. According to the HighField AI’s website, Cortex (the AI orchestration platform for graphics) is capable of the following:
“[Cortex] ingests data from newsroom systems, media libraries, and live feeds, then dispatches tasks across specialized AI agents for summarization, asset matching, compliance checks, and layout rendering. These agents work in parallel under Cortex’s orchestration logic to populate templates with verified content.”
News is a sector that has come under incredible pressure in recent years due to increasing competition from social media, which is why companies might be more incentivized to experiment with new ways to cut costs. This may be the rationale behind HighField AI’s launch.
HighField AI was also involved in the 2025 IBC Accelerator AI Assistance Agents in Live Production (I highly recommend watching this session to get a deeper understanding of Agentic AI in action).
Midstream: content discovery and monitoring
Content management shows much more significant Agentic AI development, particularly in content discovery for both production and distribution.
In July, Moments Lab announced that it would launch Discovery Agent, a tool to assist content production teams, at IBC 2025. The company had previously secured $24m funding in June to focus on developing Agentic AI technology. Co-founder and CEO Phil Petitpont said of the Discovery Agent in July:
“You can think of our Discovery Agent as your personal research assistant, one who can remember every single video in your archive… Just tell the agent what you're looking for and it instantly returns the best selections from your video library”
A photo released alongside the announcement shows users interacting with the tool through natural language, signalling a focus on accessibility for non-technical users and a new way to search for content. This was further highlighted in September announcement of the product:
“In just the past year, the way we find and consume information has shifted dramatically. Instead of scrolling through pages of search engine results we’ve started to talk to AI assistants. Tools like ChatGPT, Perplexity, and Claude help us ask questions, explore topics, and get direct, relevant answers. Search is no longer about keywords—it’s conversational, contextual, and powered by intelligent reasoning.”
This is a product at the border between content production and management: it helps creative teams do their jobs more efficiently while leveraging input from content management systems (I decided to locate the product in this section for the latter reason).
In September, ThinkAnalytics announced that it would launch a ThinkMetadataAI at IBC 2025. While this tool also focuses on content discovery, its main objectives are increasing viewer personalization and engagement, which are closer to content distribution. The company explained:
“ThinkMetadataAI uses Agentic AI to automate the process of creating enriched metadata for entire content catalogs without compromising quality. With ThinkMetadataAI, the company has solved one of today’s video providers’ biggest challenges: how to increase viewer engagement.”
What is interesting to me is that both products address the same challenge. This is making sense of the massive surge in content production over the past decade, making it more discoverable and therefore actionable for media businesses.
Aside from content discovery, Witbe announced in July that it would launch its Agentic AI solution at IBC 2025. This is focused on QA monitoring and testing. The company explained:
“Witbe Agentic AI builds on Witbe’s proven technology by introducing adaptive, goal-oriented agents that make workflows smarter, more flexible, and easier to scale.”
Content monitoring represents another promising area for Agentic AI development. The technology is suited to monitoring tasks, but it will be important to see how media technology suppliers find the right balance between agent-driven automation and human-in-the-loop oversight. Agent-driven content monitoring solutions also share some similarities with those used in distribution, which is the next focus of this analysis.
Downstream: advertising and user experience
Content distribution is also a hot area of application for Agentic AI technology, particularly in the areas of advertising and user experience.
Let’s start with advertising. In June 2025, ad tech company Seedtag announced the launch of a contextual ad targeting solution called “neuro-contextual advertising.” According to the announcement:
“Built on the combination of artificial intelligence and neuroscience principles, Seedtag said its AI agent, Liz, is now capable of interpreting deeper consumer interests and signals like interest, emotion and intent and then turning these insights into high-performing campaigns across premium connected TV, video and the open web.”
The approach stands out for its adaptive nature, dynamically tailoring advertising to consumer signals. While details on how Liz detects and interprets interests and emotions remain limited, the focus on real-time responsiveness to audience signals makes this an interesting innovation (consistent with others in this analysis).
In July, smartclip (part of RTL Group) unveiled Sidekicks, an Agentic AI ad platform developed with partner Realytics. The platform combines Realytics’ media intelligence with smartclip’s process and workflow automation:
“The platform supports operational teams with a portfolio of focused AI agents — ‘sidekicks’ — that assist across the digital advertising value chain, all while ensuring that proprietary business knowledge stays securely within the organisation.”
This is another ad-focused application, showing that advertising has the potential to be a significant development front for Agentic AI technology in media.
Another hot area in content distribution is user experience. Here, media technology providers are rolling out suites of AI-powered agents that dynamically adapt configurations (and actions) to both business goals and consumer behaviour (similar to what Seedtag is doing in advertising).
In July, ContentWise launched Agent Engine:
“ContentWise’s new AI agents are autonomous software units designed to perform tasks, make decisions, and interact with the digital environment based on user-defined goals. Integrated directly into the ContentWise UX Engine, they connect to a range of external tools and services, empowering teams to move from manual configuration to automated, goal-oriented, multi-step instructions.”
This is an interesting technology. It is a multi-agent technology and is connected to business goals like “promote trending shows to the right audience,” as mentioned by the article. Paolo Cremonesi, CTO of ContentWise, said of the technology:
“Our agent architecture, built on open standards like MCP and A2A, is designed to break down data silos. We’re moving beyond simple API calls to true orchestration, where agents can intelligently chain together tasks across multiple platforms... It’s not just about automating a single task; it’s about designing and executing complex, multi-step workflows that can pull from social media trends, trigger a campaign in the UX Engine, and push a notification through an external service…”
This points to a future where UX decisions like promoting a specific title are driven by real-time signals like social media activity.
In August, Accedo announced the launch of Accedo Compose, AI agent-powered orchestration layer developed with Merapar. As their example illustrates:
“For example, a churn risk agent can detect declining engagement and trigger retention flows tailored to individual behavior. A UX agent can adjust layout and messaging dynamically based on context, such as time of day or frustration signals. A monetization agent might surface upsell offers precisely when users are most likely to convert.”
Like ContentWise, Accedo is offering a suite of responsive agents that continuously adjust experiences based on behavioural signals. Moreover, both embed big technologies companies’ models into media-specific platforms: ContentWise builds on MCP and A2A, while Accedo leverages Amazon SageMaker Unified Studio and Amazon Bedrock.
Agentic AI hotspots
To provide a clearer and more objective view of the roundup, I’ve organized the information into a table listing all the technologies mentioned in the analysis:
The chart below summarizes the same information visually, aggregating it in content chain categories:
From this overview, it’s clear that most Agentic AI development is concentrated in content distribution, followed by content management and production. Within these areas, the primary hotspots are user experience, advertising, and content discovery.
While this is a small sample of eight products, it offers early signals of where Agentic AI technology development is headed in the media industry.
We can now move to the implications of these trends.
Implications of analysis and future directions
The following implications highlight where Agentic AI is shaping the industry most, and what it means for the future of Media Tech:
Distribution dominance: The concentration of Agentic AI in advertising and user experience shows that the industry’s priority is leveraging the technology for monetization and retention rather than for content creation efficiency (the latter is being mostly achieved through a reduction in content budgets, for now).
Content discovery remains a challenge: Despite years of investment, content discovery remains an unsolved challenge. New agent-driven solutions point to discovery as a key area, with AI expected to help bridge the gap between overwhelming content libraries and audience engagement.
From static to dynamic systems: The industry is shifting from static, rule-based workflows toward dynamic, agent-driven platforms that adapt in real time. These solutions enable greater flexibility and responsiveness to fragmented and fleeting audience attention.
Natural language as the new interface: The rise of conversational AI tools such as Moments Lab’s Discovery Agent reflects an effort to design technology products in a radically different fashion. By lowering technical barriers, these interfaces can accelerate technology adoption across organizations (and industries).
Human-in-the-loop remains essential: Content monitoring illustrates that while Agentic AI can automate repetitive tasks, media companies still need oversight to safeguard brand safety. A hybrid model of automation plus human review is likely to remain the norm in sensitive workflows.
Workforce disruption and reskilling: In many areas, agent-driven automation can already substitute repetitive tasks, and this raises hard questions about reskilling staff for the AI age (a topic I have already addressed in the previous edition of Sense the Frame).
Platform orchestration becomes a differentiator: Many solutions are positioning themselves as orchestration layers that coordinate multiple agents (mostly provided by big tech companies). Competitive advantage may therefore shift from individual agent capabilities to the platforms best able to integrate and orchestrate them.
Shorter testing and optimization cycles: This implication is consistent with the intro to this piece and, more generally, with the impact of AI on development cycles. Agentic AI enables continuous experimentation and adaptive optimization. This reduces reliance on long release cycles and shifts practices towards always-on iteration.
Technology is increasingly designed around business goals: Platforms like ContentWise’s Agent Engine and Accedo Compose show that Agentic AI is being built as a business-aligned orchestration system. Agents are configured to pursue objectives like reducing churn or promoting trending titles.
Data quality and integration become more important: Adaptive agents rely on high-quality metadata, audience signals, and content tagging scattered across different systems. Companies with richer, cleaner datasets will have an advantage, making data quality and integration even more important.
That’s all for this edition of Sense the Frame.
I hope you enjoyed my analysis. If you found it valuable, please consider sharing this article on your social media platforms. It helps spread the word and supports the time and effort I put into creating this newsletter. And if you haven’t already, I would love for you to subscribe so you don’t miss future editions.
About me
I help clients turn research and data into impact. Whether through custom reports, analytics solutions, marketing insight, thought leadership, or event content.
If you're looking for research, marketing and data analytics work, or simply want to say hello, feel free to reach out.