AI initiatives have been running for years now, and enough time has passed that businesses are recognising where they’re falling short. In this guest article, Pierre Naggar, CEO and Founder at AI automation and advisory firm BrightPro, outlines why leaders can’t simply implement AI as an add-on to existing processes, but must redesign the way work happens altogether.
If you were building an advertising business today, knowing that AI models and agents could assist you in automating tasks across your organisation, would you design the company the same way?
It is a deliberately uncomfortable question.
Advertising businesses are full of inherited complexity. Workflows have grown around client demands, agency requirements, platform changes and the constant pressure to move faster.
But the question is useful because it forces a pause. It exposes the workflows that were designed deliberately, and the ones that simply accumulated over time. It also changes the starting point of the AI conversation.
Too often, the question inside organisations is still: where can we add AI? That is understandable. A new LLM or agent framework is released, a few use cases are identified, and the business looks for places to apply it. It gives leaders something visible to point to.
But it also carries a risk. If AI is simply inserted into the way work already happens, it may improve individual tasks while leaving the operating model untouched. It may make a slow process faster, but not necessarily better.
The bigger opportunity is not to retrofit AI into inefficient workflows. It is to rethink how the work should happen in the first place.
AI Exposes the Cracks
AI is often presented as a productivity layer. Add it to the existing workflows and people will move faster. Sometimes that is true. But early AI adoption also teaches a harder lesson: it exposes the quality of the work in the organisation.
Take data readiness. I have worked on several recent projects where the ambition was to apply AI to drive more efficient sales processes. But if the CRM contains duplicates, outdated records, inconsistent job titles and unclear ownership, the data foundation is not ready to support intelligent automation.
Unless someone injects some discipline in the process and owns the definitions, rules, maintenance process and quality standards, the same problems will keep returning. AI does not fix bad data. It processes it faster and surfaces the mess more visibly.
This pattern repeats across the business. Brief-to-RFP workflows. Creative asset organisation. Campaign set-up. Reporting reconciliation. In each case, AI adoption becomes a diagnostic as much as a solution. It shows you exactly where the operating engine is broken.
The Neglected Operating Engine
The advertising industry is not short of innovation. Over the last two decades, it has reinvented how media is bought and sold. AI is now changing how planning, activation, reporting and optimisation can be supported.
But inside many organisations, the operating engine underneath the work has not evolved at the same pace. Briefs still arrive in inconsistent formats. Media plans with hundreds of lines are still coordinated through spreadsheets. Invoices are disputed because delivery numbers do not line up cleanly.
There is a human cost to this. Capable people are worn down by constant operational drag: chasing missing instructions, correcting data, rebuilding reports and keeping fragile processes moving through force of habit. That kind of work drains energy and reduces the space people have for judgment, creativity and better decision-making. In some cases I have seen it contribute directly to burnout.
The problem does not stop at the organisation’s edge. Many advertising workflows span advertisers, agencies, publishers and technology partners, which is where the dysfunction compounds. Inertia plays a part and platform fragmentation makes standardisation difficult. But the deeper issue is structural: each company chooses systems around its own needs, while the cost of those choices often lands on everyone else in the chain.
A good example of how dysfunctional this still is, is the way creative assets are communicated, a process I have seen first-hand and that has barely changed in twenty years. It creates endless friction because creative specs arrive late, assets are named inconsistently, different platforms require different formats. Nobody owns the whole workflow because it crosses company boundaries and there is little incentive to change the status-quo.
Design Around the Work That Creates Value
Most organisations approaching AI start by asking: what tasks can we automate? It is the wrong starting point because it takes the existing shape of work as a given.
A better question is: which work should be made lighter, and which work should be made more valuable?
The first category is the work that creates operational drag: chasing missing campaign instructions, rebuilding reports, reconciling delivery numbers, moving data between systems or correcting the same errors month after month.
Some of this work can be automated. Some can be redesigned. Some should simply disappear. The point is not to preserve every existing step and make it faster. The point is to ask why the work exists in the first place.
The second category is the work that strengthens the organisation’s advantage. Many advertising businesses already sit on valuable assets: proprietary data, campaign history, client knowledge, audience insight, proposal libraries, creative learnings and the judgement of experienced teams.
Take brief responses. Many advertising businesses have hundreds of past proposals containing valuable thinking: audience recommendations, case studies, performance benchmarks and lessons from previous campaigns. But that knowledge is often scattered across folders, decks and individual memories. Used well, AI can help teams access the organisation’s previous thinking and apply it to a new brief. The real value is a more relevant strategy built from what the business already knows.
The sequence matters. Cross-company problems such as inconsistent briefs, fragmented creative assets, incompatible reporting structures are real, but they require collective action and will not be solved quickly. The more productive starting point is what you can control: your own data foundations, internal workflows, naming conventions, CRM discipline and decision rights.
Fix those first. Then engage the harder inter-company conversations from a position of internal clarity rather than internal chaos.
The Real AI Opportunity
Much of the current industry conversation about AI follows a familiar pattern. Add AI to accelerate decisions. Keep humans in control. Enhance rather than replace. It is a reassuring message and not without truth. But it quietly assumes that the underlying work is worth accelerating, that the data is reliable and that the operating model underneath is fit for purpose. In many organisations, none of those assumptions hold.
Most businesses do not have the luxury of starting from scratch. But AI creates an opportunity to take a harder look at how work actually happens, and to make deliberate choices about what should change, what should stop, and where human judgement genuinely belongs.
That also requires the right ownership. Not a technology team running experiments in isolation, and not a senior leader who understands the vision but not the operational reality. What it requires is someone who combines genuine knowledge of how the business works, a clear-eyed view of what AI can and cannot do in this environment, and enough authority to drive change across functions.
The real opportunity is not retrofitting AI into every corner of the business. It is redesigning the work clearly enough that AI has somewhere useful to go and people have somewhere better to put their energy.
