Welcome to Change Makers, our series highlighting customer stories and builders making an incredible impact with Airtable. Today’s Change Maker is an Airtable MVP and community-led hackathon Grand Jury winner who built a live action game with configurable systems, persistent run history, and AI-directed entity design all on Airtable!
About the Builder
Nick Akins (
What's a fun fact most people (or your coworkers) don't know about you?
I used to rehab wild baby squirrels!
What's your superpower at work?
I think it's my desire to fully understand things. If you learn just enough to do it, it always leads to problems later, and it's always nice to be the person that people come to with questions.
What's the most surprising use case you've seen (or built) in Airtable?
Honestly, it's probably one of the submissions in this Hackathon. It's really cool to see everyone thinking outside of the box. I especially liked the Guitar Hero game.
What's a small feature or automation that makes a big difference in your day-to-day?
Two-way sync. It seems so obvious, but the ability to update across bases completely changed the way our system was set up. Even now we still find left-over corners where there's a sync and six different automations set up to try to balance it out.
What advice would you give to someone just getting started with Airtable?
Take it slow, and realize every action has an equal and opposite reaction. Youtube, Airtable Community, and Omni are all your friends!
The Challenge & Airtable Solution
Most of the time, Airtable is used to track information, not to actively run systems. That creates a kind of limitation where everything is static, predictable, and focused on organization instead of interaction. At the same time, as someone who enjoys roguelike games, Nick kept running into the same issue there too: those games can get repetitive over time because the content and variation are limited.
This project was an attempt to push against both of those constraints. Nick wanted to see if Airtable could be used as something more dynamic, and if AI could be used to introduce continuous variation instead of fixed content. On the surface, the problem it solves is boredom. But in practice, it helped him rethink how to use Airtable as something that can run systems, not just track them—a shift that's already influencing how he approaches his company's ERP build.
Nick's solution was contained entirely inside Airtable with no outside integrations. It consists of three main parts: the data tables that contain each type of record (players, enemies, bosses, game config, visual entities, etc.), the custom interface that contains the script to run the game and references the data tables, and the AI fields that create the visual and design layer of the game.
The AI implementation is where things get interesting. In the "VisualEntities" table, every entity in the game has a structured record. Things like silhouette direction, color palette, motion and animation notes, and even JSX/SVG-style implementation ideas are all generated by AI field agents. The key idea is that this isn't static. With a button press or a scheduled automation, those fields can regenerate with instructions to be as varied and unexpected as possible. That means every time you play, the game can feel visually different—almost like a new experience each run.
📺 Want to see it in action? Check out Nick’s hackathon entry here:
The Results
The biggest impact was in how much variation the system can produce. Instead of a fixed set of designs or enemies, the game can continuously generate new visual directions and ideas. That creates a system where content is not limited by how much time Nick spends building it. Each run can feel different without needing to manually create new assets every time.
It also opened up a new way of thinking about how systems can grow. Instead of scaling by adding more predefined content, it scales through controlled variation. There is no maintenance, and if you're not happy with how it looks, you can just regenerate it again. While building it, Nick started noticing that the same patterns apply directly to his day-to-day work. Using Airtable as a control surface, generating structured outputs with AI, and treating records as active inputs instead of just storage opens up a completely different way to build internal tools.
What's Next
At the moment, Nick is focused on building a fully end-to-end ERP system inside Airtable for E. F. San Juan. One of the biggest shifts happening now is moving away from third-party integrations and external automations, and replacing them with native Airtable features.
With tools like AI field agents, he's starting to rethink how much of the system can live entirely inside Airtable. The goal is to simplify the stack, reduce dependencies, and build something that is easier to control and maintain in one place.
This project also opened up new ideas around using AI in a more structured way. Instead of just generating text, Nick is interested in using AI to create controlled variation in systems, whether that's in design, configuration, or even operational workflows.
On the more experimental side, he definitely plans to keep building projects like this and participating in future hackathons! These kinds of builds are where he tends to discover new patterns that he can bring back into real-world systems.
Copy & Paste AI Prompt
Most of Nick’s prompts for this build are quite long, but a quick example is his AI Motion Pass. “The hard part of having randomly generated AI art direction is to make it still seem coherent and intentional, that is what I aimed to do with this prompt,” he says.
"You are designing motion for a small, top-down action-game entity. Create readable, efficient, and expressive motion for the core gameplay sequences. Motions must remain visually distinct at a small sprite size and be performant for real-time gameplay.
Task description:
Analyze the entity brief, its type and theme, silhouette requirements, and available animation notes. For each of these actions—Idle, Move, Attack, Hit reaction, Looping FX—write exactly one clear motion plan per line. Keep to plain text, using no extra explanations, commentary, or formatting—just direct, succinct motion descriptions for each state. If you are missing data, state explicitly which action(s) can't be described.
Output format:
Return only these labeled lines, each followed by the action motion:
Idle:
Move:
Attack:
Hit reaction:
Looping FX:
If you cannot describe a specific action, write "[Action]: Unable to provide description with available data."
Example:
Idle: Slow vertical bob, subtle shadow pulse.
Move: Quick sliding motion with squash-stretch body.
Attack: Fast angular lunge with trail accent.
Hit reaction: Sharp shrink and flash core color.
Looping FX: Faint orbiting glow particles.
Context and Data:
Name:
Entity Type:
Theme Prompt:
Silhouette Notes:
Animation Notes:
Did this use case inspire you? Drop a comment below if you're similarly using Airtable to leverage customer feedback into proactive data, and share how you're using Airtable to change the way you work for a chance to be our next Change Maker spotlight!
