Emergent Media's AI Agents: Can Marketers Replace Humans?

UAE agency Emergent Media says AI's agentic system could replace routine marketing roles. Its Emergent OS automates social, web optimization and ad buying—sparking debate over AI vs human creativity. For nearly two decades, agencies hired media planners, campaign managers, SEO specialists, analysts and community managers to navigate increasingly complex advertising systems. Brands paid for expertise in Google's search algorithms, Meta's advertising engine and the performance dashboards accompanying each campaign. Then artificial intelligence arrived. At first, it wrote copy, generated images and summarised reports. Soon it was creating advertising assets, analysing customer journeys and producing campaign recommendations. Now, startups believe it can replace entire operational functions within marketing departments. One is Emergent Media, a UAE-based digital transformation and marketing agency founded by Roshan Ejaz, Ahsan Khan, Faisal Ayub and Khurram Saleem, with a Lahore office. Its Emergent OS, launched last month, is an agentic operating system comprising three AI agents designed to automate social media management, website optimisation and digital media buying. Unlike chatbots that respond to prompts, these agents are intended to observe, analyse and execute continuously, with humans supervising final decisions. The proposition divides the industry. Supporters say machines can handle repetitive, data-heavy work faster and more consistently; critics contend that marketing still depends on human judgement, creativity and relationships. The debate extends beyond one startup: can AI genuinely become a digital marketer, or merely a sophisticated assistant? Marketing's commoditisation problem Roshan Ejaz, co-founder and CEO of Emergent Media, traces the company's journey back to a structural shift in digital advertising itself. 'Over time, as digital media evolved, we realised things were becoming very commoditised,' he says. As campaigns moved in-house and advertising platforms became more sophisticated, running them ceased to be a specialised agency capability. Agencies once differentiated themselves through media-planning expertise, and later through audience targeting. Large groups invested heavily in specialists who could extract marginal efficiencies from Facebook and Google. Meta's machine-learning systems now assess not only audience targeting but creative diversity, limiting reach when advertisements appear too similar. 'The amount of content required to achieve efficiency became enormous,' Ejaz explains. Traditional production workflows could not keep pace with platforms demanding hundreds of creative variations tailored for increasingly segmented audiences. Agencies struggled with cost, turnaround times and scale. Emergent first established a generative AI studio to produce content at scale. But people still had to transfer information between tools, analyse dashboards and perform routine tasks. The company therefore began building autonomous agents to execute marketing workflows. 'The world is moving towards automation,' Ejaz says. 'We want humans to spend their time on more intelligent and strategic work rather than routine operational tasks. We believe agents can perform those operational tasks better, while humans should focus on strategy.' That philosophy underpins all three of Emergent's AI agents. The question is whether each can actually deliver. The first, called Engage, attempts to become an organisation's social media manager. The second, Webcare, oversees website optimisation, technical SEO and analytics. The third, Boost, manages digital advertising campaigns across Meta and Google's advertising ecosystem. Each standalone agent costs roughly Rs100,000 a month, though clients can subscribe to the full operating system. Customers include brands building in-house capabilities and agencies automating parts of their workflows. Whether the agents can automate work now performed by thousands of professionals depends on the function. Engage: Can AI become your social media manager? Community management has quietly become one of digital marketing's most labour-intensive functions. Large brands maintain dedicated teams responsible for publishing content, responding to customer queries, handling complaints and keeping dozens of social channels active throughout the day. According to Ejaz, much of that workload consists of repeatable processes rather than creative decision-making. 'The Engage Agent performs all of those tasks,' he says. Users specify the content they want; the platform generates and schedules it, monitors conversations and responds in real time. Companies can upload proprietary knowledge bases covering products, policies and brand guidelines. In theory, it functions like a social media manager who never sleeps. The attraction is obvious. For organisations operating across multiple time zones, or simply handling thousands of customer interactions every day, the economics begin to change rapidly. A single AI agent can theoretically manage volumes that would otherwise require several employees working in shifts. Here, however, an experienced media executive who requested anonymity sees AI's greatest limitation. 'At the end of the day, it's responding based on prompts,' he says. 'If somebody leaves an abusive comment, the AI will probably reply with something polite like, 'Thank you for your feedback. We respect your point of view.' Anyone reading that immediately knows it wasn't written by a human.' If someone mocked his company's branding, he says, a human community manager might respond with wit, humour or cultural context, something spontaneous enough to reshape the conversation rather than simply acknowledge it. 'That's the kind of answer a human gives,' he says. 'A machine can't think like that.' But his criticism extends beyond tone. Modern social media management, he argues, has evolved into customer relationship management. If a customer complains about receiving a defective product, merely apologising is not enough. The response often requires refunds, escalation to customer service or commercial discretion. 'Engagement isn't just replying to comments,' he says. 'It's customer relationship management in the modern era. That's one area where I'd never compromise.' Ejaz acknowledges that concern to some extent. Unlike fully autonomous AI systems, Emergent intentionally keeps humans involved in critical decisions. Knowledge bases can be customised, workflows configured and responses supervised where necessary. The company's objective, he argues, is not to remove humans from communication altogether but to eliminate repetitive operational work while allowing marketers to intervene where judgement matters. Between automation and originality Umair, a digital marketing executive who has spent nearly 15 years managing campaigns across the UAE and Qatar, including work on Qatar Tourism's FIFA World Cup campaigns, occupies a more nuanced middle ground. He agrees that AI can substantially improve social media operations but not because it suddenly possesses human creativity. Instead, he believes AI excels at processing information. He points to platforms like Brandwatch, which already monitor conversations across websites and social networks before summarising public sentiment into digestible reports. Generative AI simply makes those insights dramatically easier to understand. 'Instead of reading thousands of posts yourself, AI gives you a concise overview of what people are saying,' he explains. The same principle applies to content production. Today's AI systems can generate social posts, advertising copy and creative assets at remarkable speed. Speed, however, is not synonymous with originality. 'One of AI's biggest strengths is speed,' Umair says. 'But that speed also comes with mediocrity.' He argues that AI fundamentally learns from historical patterns rather than genuine invention. It can remix successful ideas, imitate styles and identify what worked previously. But producing genuinely unexpected creative breakthroughs remains difficult. He illustrates the point through famous South Asian advertising slogans such as Tedha Hai Par Mera Hai, campaigns whose success depended precisely on being strange, surprising and culturally resonant. 'Those random flashes of creativity are difficult for AI,' he says. 'I don't think AI will ever completely replicate that level of human originality.' In his view, AI is best understood not as a replacement for creative professionals but as a facilitator. 'It can probably perform my routine daily work even better than I can,' he says. 'But when I need to invent something completely new, when I have to think creatively, that's where AI still struggles.' Webcare: Teaching websites to fix themselves Search engine optimisation has long been one of digital marketing's least glamorous disciplines. Unlike advertising campaigns or viral social media posts, SEO is painstaking, repetitive work. Specialists spend countless hours monitoring website performance, identifying broken links, analysing search rankings, checking Core Web Vitals, studying Google Search Console data and implementing dozens of technical fixes that users never notice but search engines do. Much of that work, Emergent believes, is perfectly suited for machines. Its Webcare Agent begins by analysing a company's website and integrating directly with popular content management systems such as WordPress and Shopify. Instead of merely identifying technical problems, it can recommend, and with human approval, implement changes directly within the website itself. It continuously monitors Google Search Console signals, detects ranking declines, diagnoses website issues and proposes corrective action before executing any modifications. The second layer is conversational analytics. Instead of searching through menus inside Google Analytics, a marketing manager can simply ask the system why traffic declined last week, which acquisition channels are converting best or where users abandon the customer journey. 'The analytics become conversational,' Ejaz says. 'You can ask the agent any kind of question about your data.' Unlike the Engage Agent, Webcare attracts far less disagreement from industry observers. For Umair, SEO represents one of AI's strongest use cases precisely because it revolves around pattern recognition rather than creativity. 'All of those tasks are pattern-driven,' he says. 'AI mainly helps by telling you: here's what's wrong, and here's how you should fix it. It can even generate the code or instructions needed to make those fixes.' The media executive reaches almost the same conclusion. 'That's perfectly fine,' he says of AI-powered SEO tools. 'For reporting, AI is very good.' Advertising's invisible machine The most commercially significant component of Emergent OS may also be the least visible. Known as the Boost Agent, it sits behind advertising campaigns running on Google, YouTube, Facebook and Instagram. Its purpose is deceptively simple: watch campaigns continuously, detect when performance deteriorates, recommend budget changes, pause ineffective advertisements, increase spending on high-performing audiences, create new campaigns and optimise bidding. According to Ejaz, this replaces one of the most routine jobs inside every digital agency. 'The person sitting inside an agency or on the client side constantly monitors campaigns,' he says. 'They're increasing budgets, decreasing budgets, pausing ads, checking performance. All of that work is now handled by this agent.' Importantly, Emergent stops short of granting complete autonomy. Every optimisation still requires human approval. The company deliberately keeps 'a human in the loop', even though the platform connects to multiple language models including Claude, OpenAI and Gemini. 'Technology can fail at any time,' Ejaz says. 'Otherwise something could go wrong, your client's budget could be spent very quickly and nobody would realise it.' Understanding whether Boost represents genuine innovation requires understanding how digital advertising actually works. Umair says public discussion of AI in media buying often misunderstands the industry. 'There is no person sitting there negotiating prices,' he says. 'It's all based on bidding.' Every second, Google's and Meta's advertising platforms conduct automated auctions, evaluating thousands of competing advertisers attempting to reach similar audiences. 'The bidding is done by machines. Every advertising platform has its own machine performing real-time bidding.' Artificial intelligence does not replace Google's auction. Instead, it operates one level above it. 'I can tell my AI never to bid more than two dollars,' Umair explains. 'The AI studies historical data and decides which audiences deserve more spending and which audiences should receive less.' So who loses their job? If Emergent's technology performs as advertised, its biggest impact may not be on marketing budgets or advertising performance. It may be on payroll. Ejaz does not present Emergent OS as a replacement for marketing leaders or creative directors. His vision is considerably narrower, but still disruptive. Umair goes further. 'If someone's entire job is producing reports, exporting Excel sheets, building pivot tables and presenting numbers, that person is highly replaceable.' He includes SEO specialists, digital media planners, media buyers and analysts whose work revolves around repetitive optimisation. 'SEO is repetitive. Media planning is repetitive. Data analysis is repetitive. Those jobs are primarily about identifying patterns.' The media executive accepts much of that argument. He readily acknowledges that AI is extraordinarily good at reporting, campaign monitoring and identifying optimisation opportunities. 'Where four people are currently doing the work, one AI agent can do the work of all four,' he says. 'It saves time, saves energy and it's accurate.' His disagreement begins when optimisation becomes decision-making. Modern advertising, he argues, often requires judgement that cannot easily be reduced to mathematical objectives. 'The AI will either chase CPM or chase reach. It won't be able to pursue both simultaneously because it isn't human.' 'If a human has to monitor it, then we're back to the same thing,' he says. 'You paid money to build the bot, and then you also hired a human to sit there and monitor it. Then what's the benefit?'