Writing

What would it take to design a prompt configuration that runs Sanju Samson? A system prompt for identity. Macro context for the innings. Micro context for every ball. And a priority tree that resolves when layers disagree.

Read the full piece →

And your projects. Just today I was trying to remember the last time I enjoyed breakfast. The last time I wasn't rushing to the next thing. The last time I actually relished a meal.

And just today I caught myself rushing through a design problem on bowldem. Wanting to ship it, move on. But why?

I need to chew it.

That's how you enjoy projects.

The AI-native game space is still very early - hour one, if you will. But there's a big change coming, and the signs are there. Mobile gaming especially has been lagging behind, which is unusual because gaming typically leads innovation at the intersection of art, entertainment, tech, and consumer behavior.

What Makes a Game AI-Native?

There's confusion about what AI-native actually means. People often conflate it with using AI in the development pipeline - procedural generation, asset creation, or level design.

A simpler way to think about it: If you remove AI from an AI-native game, the game collapses.

AI needs to be at the center - driving the core experience and moving the player from point A to point B. The player must feel the AI. They don't need to know whether AI is generating the response, simulating logic, or mimicking a player - but they should feel the presence of intelligence in every interaction.

The PC Pioneers

AI2U: With You Til The End - the player interacts with a Yandere girl while trying to escape. Every behavior - whether she leaves the room, stays, or offers hints - is driven by AI. Remove AI, and the entire experience collapses.

Suck Up! - another strong example. You play as a vampire negotiating entry into homes or talking to police officers. The entire progression depends on AI-driven dialogue and dynamic social reasoning.

The Companion App Wave

Ani and Valentine from Grok have set new standards for AI-driven companions. Their responses are rich and non-deterministic - but more importantly, AI controls emotion, movement, expression, and environmental cues in real time.

Mobile Gaming: The Add-On Problem

In most cases, AI shows up as an add-on feature. You complete a level built by code and logic, and then hand over the experience to AI for a chat or report card. The core loop remains traditional; the AI layer sits on top.

The next wave needs to move beyond enhancement into AI as the core interaction layer - where the gameplay itself wouldn't exist without AI.

Why We're Excited About Building AI-Native

At Ripple AI, we're building at the edge of what's possible. It feels like hour one again. Things are only going to accelerate from here.

The opportunity is massive. LLMs are becoming more capable. Hardware is improving. Infrastructure is ready. Everything is aligning.

Because that's where this is headed. Not AI-assisted. Not AI-enhanced. AI-native.

We tried to prove it wrong. A 6-week experiment.

Terra had 50+ games. Sports, action, racing, casual. Players would jump from Cricket to Tennis to Olympics, but there was no platform-level reason to come back tomorrow. We had Terra Coins, basically Robux for our platform, earned through daily streaks and spendable on in-game boosts.

If every successful game needs progression, shouldn't platforms need it too? So we built one. Six weeks to prove it mattered.

The Bet

Games use XP to keep players coming back. Some platforms tried it. Steam for badges, Epic for achievements, Google Play for cross-game progress. But we didn't know if players actually cared about platform progression.

What Happened

We added a level-up tree. 16% of players clicked through from that page into games. For a meta-layer feature, that's real engagement.

Entry into favorite games went up. Items beat Terra Coins on gametime.

But retention barely moved. D1 was slightly better for TC. D2, same story. D7 and beyond, no meaningful difference from control.

The system worked. It just didn't matter.

What We Got Wrong

Progression works when it's the point, not when it's decoration. Players don't need meta-progression to enjoy a gaming platform. They need good games and easy access.

Progression systems for platforms? Overrated.

Several years ago in a meeting room that smelled like it had been occupied for way too long, we discussed our next bold beat. Bold beat - a term from gaming for a major feature release.

I lost it internally.

New bold beat? We were struggling with regular releases. Every release had a hotfix. The team worked weekends just to cobble things together.

The Wrong Optimization

On one hand: you need stability. Clean releases. Predictable timelines. This is responsible leadership.

On the other hand: time is precious and every game is at war to demand player attention. Test fast, fail fast, ship fast. Six months to ship anything is a death sentence.

My mind was wired for the first. The market demands the second. It didn't even register.

I sat there thinking everyone else was being reckless. I was the reckless one. Reckless with time. Reckless with opportunity. Reckless by being risk averse. Craving stability that ensures death.

Jeff Bezos provided a framework that I found interesting. He was a master of inverting.

"I very frequently get the question: 'What's going to change in the next 10 years?' I almost never get the question: 'What's not going to change in the next 10 years?' And I submit to you that that second question is actually the more important of the two."

In game development, the pace of innovation is extremely fast. If everything else is changing, a good question to ask yourself as a game producer is - "What will not change in the next decade?"

It is impossible to imagine a scenario where great games and software updates will not be necessary to be delivered with great urgency and high predictability. Time to market will make or break game studios.

Your game development strategy needs to be focused on not speed, but instead velocity. Your entire organization needs to be set up that way. You need to be ok to take hard calls. You should be ok to prioritize ruthlessly. You must be able to provide focus for your team. Don't add, but instead subtract.

Focusing on velocity ensures that the team is moving in the right direction. Focusing on volume ensures you have more frequent updates and more releases, but overall you will find you are shipping less meaningful work.

What do I stand for? Through introspection and reviewing professional and personal relationships over the last couple of decades, I've identified three core values.

Integrity

A long, long time ago I had an interesting conversation with one of my professors during a surprise hostel raid. He found a bottle of Old Monk in my cupboard and offered me an out - "Tell me whose bottle this is?" My reply was quick - "This bottle is mine."

I don't cook up random stories. I am honest and show a consistent adherence to moral and ethical values, however uncomfortable it might be.

Respect

I take a long term view of relationships. "It has been said that astronomy is a humbling and character-building experience. To me, it underscores our responsibility to deal more kindly with one another."

I like to treat people the same way I would like to be treated - with respect. Being respectful also means being honest.

Resilience

Failures are quite common in game development. There are more chances of a game not working than of an outsized success. I have made peace with failures. I try now to instill the same in the teams I work with.

January 2025 Edit: The world is my Oyster. "Everything around you that you call life was made up by people that were no smarter than you."

Finding the balance between creativity and delivering a game on time is hard, but necessary. It's probably the most important trait a producer needs in game development.

The best leaders I have worked with typically approach this problem with the following mindset:

• There needs to be periods where the team needs to be creative. During this period, they have the license to go completely blue sky & off the rails.
• There will also be periods where the team will focus on execution. Heads down & build.

The rhythm of moving game development along the phases of divergent and convergent thinking will eventually determine the success of your game.

Another useful way to think about creative decisions: is it a two-way decision or an irreversible one? Take plenty of time for the irreversible decision points. For the ones you can walk back from, no point in mulling over it. Pick a solution and move as fast as you can.

I was an athlete once.

I loved sprinting. My choice event was the 100m dash. I ranked among the top three in my school. I built quite a collection of bronze and silver medals.

But I wasn't the fastest.

A week before a sports tournament, the coach shared what events we would participate in. I was sure my name would be picked for the 100m dash. My name only made it to the 1K. I didn't even know that there was an event like that.

Come race day, I found myself doing exceptionally well running long distances. I won my first 1K and finished 2nd in the finals.

The coach saw then what I didn't. I didn't have it in me to be the best at 100m dash. He took my talent, and he maximized my chances of success by training me on stamina.

He broke me everyday, but he picked the right horse for the course.

For over a decade, I've been playing Fantasy Premier League (FPL). And for over a decade, I've been consistently terrible at it. So I decided to build something to help analyze it better.

One small catch: I hadn't coded in years.

First Attempt: The Screenshot Dream

I dove into Optical Character Recognition (OCR). Teaching a computer to read player names from screenshots is about as reliable as my captain picks. Sometimes it would read "Mohamed Salah" as "No Bad Sala".

The API Pivot

Once I started working with the FPL API, things clicked. My first successful API call felt like scoring in the 90th minute!

The Final Product

A web application that rates your FPL team on a scale of 1-100, analyzes player performance based on position, and considers upcoming fixtures.

I'm still not great at FPL, but at least now I have data to explain why.

Try it here

IGDC 2020

An interactive session on how I think of Production as 4 key pillars across Project, Process, People and Player Management.

I join Stan Minasov from AppMagic in the 9th episode of Games & Names. We cover everything LiveOps - challenges, pitfalls, and the secret sauce of maintaining a performing game for years.