Collaborative intelligence: How humans and Generative AI are achieving more together

In a short span of 18 months, Generative AI has leapt from party-trick chatbots to work-ready co-workers thanks to a string of recent breakthroughs, such as reasoning ability, real-time online research, vast context windows, and true multimodal understanding.

At OMG, we are at the forefront of harnessing these breakthroughs. Our strategic enterprise partnerships, including a first-mover collaboration with OpenAI in 2023, granted us early advantages, enabling us to develop cutting-edge solutions for our clients and transform our own operational efficiencies.

These broader AI advances are game-changing because they transform AI into a more practical tool that can integrate directly into real job tasks, such as analysing documents, drafting creatives, summarising hours of conversation, and even critiquing what a camera sees in real-time. 

1. Chain of thought (CoT) reasoning in new LLMs 

Modern Large Language Models (LLMs) such as Open AI’s O3 & Gemini 2.5 Pro employ reasoning to deconstruct complex problems into a sequence of logical steps, a method mimicking human cognitive problem-solving. 

It enables these models to internally understand context, draw logical conclusions, and solve complex problems, making their responses more accurate and reliable across real-world applications. This has drastically improved our social listening capabilities for brands such as Mead Johnson and McDonald’s Hong Kong, enabling us to interpret brand sentiments, local nuances, and context that traditional methods might miss. 

2. Real-time research and validation 

Through the combination of Retrieval-Augmented Generation (RAG) and new specialised web search capabilities, models such as GPT-4o can pull the latest facts, cite sources (internally or externally) and cross-check claims on the go. 

Fresh data turns AI from a static knowledge base into a living researcher. For one of our clients – a leading bank in Hong Kong – we are prompting Gen AI to dissect landing page content, evaluate its unique selling points, and instantly generate brand compliant, high-performing search ad copies in multiple languages. All these have driven significant improvement in click-through-rates. 

3. Giant context windows 

Leading models now ingest up to millions of tokens in a single prompt, the equivalent of an entire book’s worth of words. When you can feed an AI weeks of social content or a year of call centre transcripts, it can surface hidden patterns no spreadsheet could reveal.  

By leveraging this advancement, OMG Hong Kong can load large volumes of public social conversations into the model to receive concise, theme-clustered insights that steer marketing strategy in hours rather than weeks. A successful example is our test with Mead Johnson Nutrition, where we saw significant improvement in ad engagement after using AI-generated insights from our proprietary Gemini-powered social intelligence hub to alter creative angles, process ambiguity, colloquial idioms, and complicated sentence structures. 

4. Multimodal intelligence 

Models such as Gemini Live and GPT-4o process text, images, video and audio interchangeably, producing responses that feel closer to human dialogue and experience, grounded in what the camera or screen is “seeing.” 

Seamless real-time understanding of mixed media unlocks creative analysis at scale. The use of multi-modal AI allows us to examine our clients’ display, video and social ads side by side to identify overarching creative directions and opportunities for differentiation. This technology was used for one of our FMCG clients to identify their competitors’ ad investment split between branding and conversion-driven creatives. While AI still has room for improvement in certain specialised, high-stakes, or physical-world tasks, the pace of progress means that teams that prepare now will be ready when the next breakthrough arrives. 

Shifting the narrative: From job uncertainty to productivity boom 

Rather than dwelling on whether Gen AI will cause job uncertainty, we should focus on its potential to revolutionise how we work and unlock unprecedented efficiencies. Here’s how: 

I. Amplifying human potential, not replacing it  

Generative models can handle the repeatable and data-dense tasks, freeing humans to focus on strategy and relationship-building. The net effect is a productivity gain, not redundancy.

II. Domain expertise elevates AI

A large language model can draft a media plan, but only a domain expert knows the client nuance, the regulatory barriers and the brand tone that makes the plan viable. Human judgment transforms generic AI output into business value. In fact, several advanced AI capabilities have been incorporated into Omni, Omnicom’s open operating system, to streamline and enhance media planning workflow. These include audience and market understanding, budget optimisation, and reporting automations.  

III. Prompt engineering = Digital literacy 2.0 

Without clear and well-crafted instructions, AI is merely potential waiting to happen. Teams that learn to decompose problems, feed the right context and iterate on prompts will outpace those who wait for "one-click magic".

IV. Build data foundation and AI workflows 

True transformation demands clean, well-governed data pipelines and streamlined processes: where data lives, how it’s tagged, who approves outputs and how results loop back for continual learning. Companies that invest here will convert AI curiosity into durable competitive advantage. 

How brands can lead the change 

The pivotal question now is not if Gen AI will impact your business, but how you will lead that change. It's vital to recognise that even if current AI capabilities have room for improvement, the technology is evolving at an unprecedented speed.  

To unlock Gen AI's potential, companies must actively explore its applications. This transformation hinges on the willing engagement of all departments. By fostering this collaborative approach guided by human expertise, streamlined workflows, and solid data foundation, businesses can achieve greater efficiency and better outcomes.  

Furthermore, this effort must rest on a strong commitment to deploy AI responsibly, upholding the highest standards of data security and privacy to ensure that progress builds trust. At OMG, we are harnessing AI to drive efficiencies and effectiveness in media, focusing on responsible innovation rather than chasing hype. We are dedicated to building use cases that deliver real value and improve ROI for our clients and consumers. The future remains human with AI, enabling us to achieve more together than either could alone. 

This article was written by Chris Kwok, data marketing director, Omnicom Media Group Hong Kong.