What AI can and can not do for your agency

Learning the potentials and limitations of Generative AI

Anna Legros, Strategy Director Ogilvy, and Joe Procter, Managing Partner, Open Partners compare notes on how artificial intelligence can change the way agencies work – or not.

How I look at AI

    Joe Procter: Magic needs more than AI

    At Open Partners we embarked on a mission five years ago to embrace change and enhance marketing effectiveness for our clients. Along the way, I've adopted the mindset of a Paranoid Optimist - a blend of caution and hope - as we position our clients to win across Media, Creative, and Technology.

    In those five years, nothing has come close to the disruptive force of AI.

    Productivity gives reasons for optimism

    On the one hand, it offers abundant reasons for optimism, not least when it comes to setting agencies like ours apart from the competition. The way we use AI versus our peers has suddenly become a key differentiator, and our Automation team has been deployed for some time to interrogate our business, develop specific use cases, and demonstrate how we use automation - including AI - to produce better, more effective work for our clients, for less.

    But when it comes to AI specifically, there’s no ivory tower at OP, and it’s definitely not just the responsibility of the Automation team. Given AI’s capability to supercharge our entire operation, we’re so optimistic that we're making it everyone's responsibility. Prompt Engineering training is being rolled out across the business, equipping each of our 106 staff to be fluent in AI, guaranteeing effective communication between our Human Resources and our new ‘Machine Resources’, and ensuring that communication with our new AI co-pilots is on par with our communication with our media partners, clients, and their customers.

    Early achievements so far have been great, especially for tasks that don't require us to trust the outputs too much. For instance, we now utilise AI to record, transcribe, and document client meetings, providing instant access to notes and predicting actions based on what was discussed, saving us roughly two hours per meeting. Additionally, our Planning team now employs AI tools to consolidate trusted content from our desk research into concise transcripts and actionable bullet points, freeing up time to generate better ideas, create better plans, and having done it using fewer hours.

    And there’s loads more to come. Once we have the entire workforce ‘plugged in’ I’m expecting the operational, efficiency-focused use cases to flood in, at which point AI will most definitely lead to us being able to do more with less, in every area.

    So from that perspective - one based purely on productivity - my optimism is unbridled. More work. Better work. Lower costs. Happier staff. Happier clients.

    Maybe I should leave it there. I don’t want to end on a low note. But I am a Paranoid Optimist after all, and it’s time to move to the other side.

    Low-value agency tasks

    I worry most about the big stuff. Long-term stuff like the development of talent in our business, and in the industry. If AI takes over low-value agency tasks, how can entry-level roles cut their teeth? An industry where agencies are made up exclusively of senior leaders and thinkers will only last a generation or two, and then what?

    And then the big one - relinquishing the least tangible but most valuable element of our work to machines: creativity. Whilst outsourcing simple tasks is fine, ideas are different, and with limited transparency regarding AI’s sources, biases, or methods, our ability to provide the right answer - and critically to show our workings - will be lost. Human creativity has the unique power to draw from and create culture, whereas AI can only mimic it based on its training.

    So in summary, for me AI is like a dazzlingly bright new colleague, brimming with potential and eager to get cracking on everything, but lacking emotional intelligence, and almost certainly untrustworthy.

    Joe Procter, Managing Partner, Open Partners

    I’m highly optimistic about its ability to help us do existing things faster (and be more agile as a result), but when it comes to the ideas that will transform our world again, I’m worried that outsourcing these to a machine will result in a bad ending for us all.

    So for Open Partners, I intend to use AI to eliminate the grind, to give ourselves a head start in building the foundations of our ideas, but for us to always be in control of the magic.

    Anna Legros: An opportunity to swim deeper, not just faster

    Autumn 1997, I landed my first job, at a London digital agency. An arts graduate, I was sent to an internet café on Tottenham Court Road for instruction on how to use search engines. We’re talking Ask Jeeves and Altavista for Yahoo, with a little early Google. I learned how to create basic search terms and get a response from a limited pool of online information.

    This memory revived when I took a prompt engineering course for ChatGPT.

    Today, learning the potential/limitations of Generative AI is essential. Not only for the day's task at hand but for how we set our ambition.

    Most of us aren't new to AI. If you've worked with creative technology over the last 8 - 10 years, your campaigns have likely used it. Ogilvy was the first to use Generative AI tool DALL.E in advertising for Nestlé’s La Laitière. Our data strategy colleagues have been in the thick of it for a while.

    Democratising our use of the technology

    It's a direct access to the tsunami of new Generative AI tools that is democratising our use of the technology. And that has quickly made it the responsibility of every agency and brand discipline.

    What does that mean for Strategy? We've seen the charts that show the speed of technology evolution has never been this fast before and will never be this slow again. The velocity and depth of impact described as a ‘cultural’ revolution, not a technology revolution.

    We've been working faster for the last couple of years. Does faster mean better? Can Generative AI mean better?

    It gets us quick access to frameworks, explainers, vocabulary, ideas for ways in, diverse voices, and stimulus. Voice to text to styled formats, data visualisations, cultural stimulus.... One senior strategist confided that, ChatGPT can be a source of bursts of confidence, validating existing thinking, in a world of no single right answer.

    Did it save me when I had 30 minutes to find examples of the UK’s attitude to being naked? Yep. Including a lovely YouGov stat. Did a new Generative AI app touting itself as an alternative to a creative agency produce original and distinctive concepts? Not for now.

    Real strategic value that Generative AI brings

    Is this just about speed? How can we learn about the real strategic value that Generative AI brings to our work?

    We need to be serious about playing with Generative AI technology, every day. Play to understand the value of precision and that the focus on the depth of our thinking is more important than ever. Discover that the results that we get are often very generic, for now. See glimpses of deeper value that will support us in our job.

    Anna Legros, Strategy Director, Ogilvy UK

    Explore cultural tensions or make complex and often contradictory information simple. Experience what helps decision-making, persuade and feed better creative experiences. See that our role doesn't change but how we do it might.

    Outputs only as good as inputs

    Crucially, we need to play to understand the need to be hyper-vigilant to bias and prejudice, embedded in the systems. Story old as time. The outputs are still only as good as the inputs, whether that is the global data sets that it uses or our framing of tasks.

    If we are setting a strategic ambition for this cultural revolution, it’s to learn to use Generative AI for good work, better work.

    I'm inspired by the power of campaigns from Ogilvy’s Deutsche Bahn No Need to Fly and Wunderman Thompson’s Rewrite Iran to the recent Asics campaign, retraining AI's prejudice with images of what sports people really look like.

    These show that the opportunity for strategists is to learn to swim deeper, not just faster.

    See more practical resources to help agencies navigate the evolving landscape of AI


    The opinions expressed here are those of the authors and were submitted in accordance with the IPA terms and conditions regarding the uploading and contribution of content to the IPA newsletters, IPA website, or other IPA media, and should not be interpreted as representing the opinion of the IPA.

    Last updated 01 May 2024