At Agoda, we love a challenge, and taking on a new technology is no exception. Welcome to Part Two of our AI in Creativity series, where Jennifer Jones (Global Brand Creative Senior Manager) shares how the Brand Creative team experiments with GenAI to boost creativity, streamline our processes, and do more with the resources we have.
In Part One of this series, I shared how my team and I embarked on our journey to explore AI in creativity. What I only briefly touched on was the mindset and approach we adopted at the outset, which became the foundation for everything we’ve achieved since.
Our goal was both clear and ambitious: to enhance the quality, consistency, and speed of our output. The idea wasn’t to use AI tools to replace creativity but to eliminate the repetitive, mundane tasks that often slow down the creative process.
Picture a workflow where AI handles 90% of the tedious work, giving you the time and mental space to focus on the most experimental and innovative aspects of your craft. That was our vision: a challenging goal, set intentionally high to push us beyond our comfort zone.
Our Philosophy for Exploring AI in Creativity
We approached this project with a simple philosophy: there are no failures, only learnings. This wasn’t just an experiment with tools; it was a chance to embrace two of Agoda’s core values: “Move fast and be efficient” and “Be a scientist: experiment and measure.”
The idea was to move quickly, test an idea, iterate, and if it didn’t work, move on to the next one. Along the way, we’d document everything—every success and every misstep— to ensure that no learning was lost. Even if a tool or approach didn’t deliver on its promise, it would still be a win, because we’d gain clarity on what didn’t work and why.
With this mindset in place, we dove into our Proof of Concept, focusing on two key areas to start with: concept and content creation and improved efficiency. Each use case began with a clear definition of the problem, our goals, and the expected outcomes. We also documented our current investment—time, effort, and resources—and set benchmarks to measure whether AI tools could deliver real value beyond being a flashy gimmick.
Successes with a Learning Curve
One of our early successes came in video ad concepts. Producing fresh, creative ideas for video ads is time-consuming, especially under tight deadlines. Our goal was to test whether AI tools like ChatGPT, Midjourney, and InVideo could generate “good enough” concepts quickly.
While early results were generic, with some ideas requiring significant refinement, we discovered that by fine-tuning prompts and creating custom GPTs, the outputs became more usable. These AI-generated concepts weren’t a replacement for our designers and copywriters but provided an excellent starting point.
Another of our standout successes was in voiceovers. Traditionally, voiceovers involve lengthy lead times and significant costs, particularly for multilingual projects. We experimented with tools like Narakeet and ElevenLabs to create high-quality voiceovers quickly and affordably.
The results were transformative: what once took days could now be done in minutes, saving up to 99% of production time for simple projects. While early outputs were robotic, ongoing adjustments improved the naturalness of the results. Today, we even have custom voices cloned for specific needs, giving us unprecedented control over tone and pronunciation.
A Mixed Bag of Results
Not all of the experiments were home runs. In our attempts to use AI for original image creation, we tested tools like Midjourney and Firefly to produce visuals for social media, ads, and mixed media projects. While these tools allowed us to create bespoke, eye-catching images, the time saved was minimal due to the refinement needed to align with our brand standards. The results were mixed—useful for creative exploration, but less impactful in terms of efficiency.
Another critical focus area was video editing, where the need to produce frequent, localized social media content presented a significant challenge. Using tools like InVideo, Canva, and CapCut, we prioritized quantity over quality, testing whether AI could streamline the creation of simple videos for various markets. The results were promising, with estimated time savings of 60% for basic projects. While these AI tools excelled at producing short, simple videos, they fell short when it came to creating content with complex effects or the polished quality achievable through traditional editing methods.
Funny Failures
One notable failure was our attempt to use AI for character creation, specifically to generate Agoda’s beloved mascot: the Agojis. Agojis were designed to embody the five dots of Agoda’s logo, with expressive eyes and limbs used to convey endless situations.
Despite trying tools like Stable Diffusion and Firefly, the outputs were inconsistent and failed to meet the exacting quality standards required for such a visible brand element. While this experiment didn’t succeed, it highlighted an important truth: not every problem has an AI solution, at least not yet. Still, the process of trying, failing, and documenting our findings proved invaluable.
Our Biggest Takeaway from Using AI in Creativity
This iterative process of testing, learning, and refining reinforced the biggest lesson of all: AI is a creative tool, not an all-in-one solution. Its effectiveness relies on thoughtful application and human oversight. By treating AI as a collaborator rather than a creator, we’ve unlocked new efficiencies and opportunities while maintaining the quality and creativity that define our work.
In Part Three of this series, I’ll explore how we applied these learnings to scale AI adoption across our workflows and unlock further efficiency gains.
About the Author
Jennifer Jones is a Global Brand Creative Senior Manager at Agoda.
