Curry on Curry
How do you translate complex AI and cloud capabilities into a highly-engaging, episodic social campaign that reached 88.7 million users and shifted global brand perception?
The Challenge of AI Perception
As Google Cloud rapidly pivoted its enterprise narrative toward the "Agentic Enterprise" and advanced AI, we faced a fundamental challenge of the AI space.
AI capability is incredibly abstract. Our strategic goal was to tangibly demonstrate the raw processing power of Gemini to a broad, mainstream audience without relying on dry technical jargon. The core KPI wasn't just views; it was a measurable lift in AI Leadership Perception – proving to enterprise decision-makers that Google Cloud is the undisputed leader in applied AI.
Building on the Human Connection
We needed a subject with enough data to push the model, and enough cultural gravity to hold an audience's attention. Stephen Curry is the most heavily analyzed shooter in the history of the NBA.
Our hypothesis: Could we use Gemini to find patterns in his game that even he didn’t know existed?
To connect with audiences and turn this data into a compelling narrative, we built on the foundations of their authentic human connection.
We paired Steph with his former Davidson College coach, Bob McKillop. By feeding our AI insights directly to Coach McKillop, we transformed a technical data-processing demo into an organic, emotional conversation built on their shared years of effort and growth.
Designing the Custom Agent
Working closely with Google Cloud engineers, our team architected and refined a custom, first-of-its-kind Gemini agent.
We built the agent to run complex, multi-layered queries across a massive dataset covering Steph’s entire 16-year career, cross-referencing it with hours of YouTube performance footage. The system was helped us surface hyper-niche insights like his exact 3-point percentage after 7+ dribbles compared to his brother Seth, or his turnaround time on no-look threes.
1,182
4,729
2,291
40,912
three-pointers
games played
miles run
minutes of court time
Production
With the data locked, the challenge shifted to physical execution. We collaborated with our external production partners, directors, and Stephen Curry’s own production team to ensure the technical premise didn't overshadow the human element on set.
Directing a shoot with high-profile talent requires absolute efficiency. Because our insights were generated and validated by our custom agent beforehand, we were able to brief Coach McKillop with precise, high-confidence statistics. Our role on set was to ensure the visual direction and the delivery of the prompts felt entirely authentic, bridging the gap between the complex data architecture running in the cloud and the authentic, human interaction happening on camera.
Agile Optimization
Once the campaign was in flight across Meta, X, and YouTube, we closely monitored the initial engagement data.
To maximize our algorithmic reach, the team drove an on-the-fly tactical pivot during the second phase of the campaign. We completely optimized the next batch of deliverables:
Format Shift: Converted standard video posts into highly engaging carousels.
Pacing: Aggressively reduced video length to spike completion rates.
Brand Camouflage: Reduced obvious corporate branding and removed the "collab" tag, which we identified was artificially suppressing algorithmic reach on social feeds.
The result of this real-time creative pivot?
176% growth in viewers reached
149% growth in total engagements
