Vishal Jacob on the rise of agentic orchestration and the road ahead

The first trend outlined in the TYNY report points to a fundamental shift in how AI is deployed in marketing — from assistive agents to scalable agentic ecosystems.

As Vishal Jacob, President - AI & Digital Solutions, Choreograph, explains, the industry is moving beyond standalone tools and isolated “agentic workflow systems” toward true agentic orchestration: interconnected networks of autonomous agents capable of managing entire marketing lifecycles with minimal human intervention. Tasks that were once linear, such as research, summarising, present — are now distributed across specialised agents that gather data, synthesise insights, quantify investment implications and translate findings into client-ready narratives.

In this model, intelligence is no longer sequential but collaborative. Agents interact with other agents, solving increasingly complex, multi-layered business problems — from deep research conducted autonomously across the web to systems that decode category shifts and model growth scenarios.

For brands, this evolution requires structural readiness. A robust data architecture that enables seamless interaction between agents will be foundational. Equally important will be specialised AI leadership, with dedicated teams or emerging roles such as “Chief Prompt Officers,” to orchestrate, govern, and extract strategic value from these ecosystems.

But as AI systems become more autonomous and layered, a critical question arises: where does human judgment fit within this expanding machine-led workflow?

Q. You’ve described a multi-agent ecosystem where AI systems collaborate to solve increasingly complex tasks. In that framework, where does human judgment remain indispensable?

Vishal Jacob (VJ): Human intervention is critical from the very beginning. An agent is designed to perform a specific task, but it needs clear direction. Humans define the task, provide the relevant data, and set the context through precise prompts. This enables the agent to interpret the ask and deliver the desired output.

In early 2025, most agents were built to handle simpler tasks such as reading and summarising documents. However, the complexity of tasks has increased significantly. Today, an AI system could be required to produce a 20-plus page research report, which involves deep research, contextual understanding, data extraction, visualisation and structured storytelling.

To address this, we have built a deep research agent, which works as a super-agent interacting with four or five specialised agents. One scrapes and identifies relevant topics, another structures the information, and a third builds a cohesive narrative. While these agents work in a coordinated manner, humans still play a key role in ensuring the right orchestration and outcomes.

Q. As agentic systems move from simple summarisation to producing multi-layered research, insights and narratives, how are you rethinking talent development to ensure teams can orchestrate these systems?

VJ: We have a robust and structured training programme supported by an internal AI community. At the start of every year, we work closely with our talent teams to create a training calendar. Classroom training remains important, but we believe a large part of learning happens through peer-to-peer collaboration.

We have identified around 40 people within the organisation to form a strong AI community in India. This group regularly collaborates to solve problem statements. When one person raises a challenge, several others contribute solutions. This collaborative learning approach has been very effective and is shaping how we conduct training today.

Q. Given this shift, how different is your hiring strategy today compared to traditional media models? What skills are becoming non-negotiable?

VJ: For Choreograph, hiring is very different from traditional media. We do not hire media planners alone. Instead, we bring in diverse skill sets required to solve client challenges.

We have strong analytics teams, so we hire data scientists who can build measurement and predictive models. We also recruit cloud engineers, which was not common in the media ecosystem earlier, as many models now run on cloud platforms.

Additionally, we look for strategists who can understand business problems and solve them using data and technology, along with product developers and UI-UX specialists. The goal is to build capabilities that allow us to consult clients on their data and technology challenges.

Q. How are client briefs evolving in an agentic AI ecosystem? Are conversations moving beyond media planning into deeper data, technology and business advisory roles? 

VJ: Media planning and buying remain core, but the nature of conversations has expanded. A significant focus today is on first-party data. Clients are asking us to enrich their existing data with additional attributes, often through privacy-safe collaboration environments such as clean rooms.

There is also increasing demand for predictive modelling. Clients want to understand potential campaign outcomes before execution. We analyse historical data, audience behaviour, creative performance and platform metrics to predict what will work.
Another emerging area is trend forecasting. Clients are now asking agencies to predict category trends, sometimes two years ahead, and even evaluate where marketing investments should go. This shifts the conversation from media planning to product and business consulting. Importantly, clients are more open than ever to having these discussions with their media partners.

Q. How would you describe Choreograph’s evolution in India so far?

VJ: It has been a strong and positive journey. We have an 85-member team in India and worked with around 35 clients last year. Most of our conversations revolve around data, technology and analytics.

We have also launched an internal operating system, WPP Open, to transform the organisation. Through Open Media Studio, we have built nearly 200 agents that enable employees to create their own solutions and improve productivity. Overall, it has been a strong year and a half for us.

Q. As you scale WPP Open and Open Media Studio, what will define the next phase of growth? 

VJ: Our priority is driving the adoption of WPP Open and Open Media Studio. We are also building something called Open Intelligence, which will power future campaigns. This involves creating three layers of models: a foundational country model, client-specific models, and activation models. Together, they will help identify the right audiences and optimise campaigns more intelligently.

We are also seeing strong demand for consulting. Many clients want guidance on selecting the right technologies and building data ecosystems. Supporting them in this journey is a key focus area.

Q. As AI infrastructure expands, how do you view sustainability in this context?

VJ: Data clean rooms are virtual and designed to ensure privacy while enabling secure data collaboration. Currently, their usage is relatively limited, and they are not deployed in every campaign. The time required for such exercises is also short, so the overall impact on energy consumption is minimal.

In comparison, large language models and generative AI applications such as video and image creation have a much higher energy footprint. While sustainability is an important consideration, clean rooms themselves are not a major contributor to carbon impact at this stage.