- Monday Thought
- Posts
- From Game Chips To Copilots
From Game Chips To Copilots
Why GPUs, game engines, and concert halls all remind us that nothing beats people in sync
We went to see Benson Boone with the kids last night. The thing about Sunday night concerts that start at least an hour late on a school night with a stadium full of schoolchildren is…
But all jokes aside, the show was fantastic. Watching live human musicians performing in the age of AI hits different. Nothing replaces the energy of people playing together in sync. Which is exactly today’s theme.
Lately I can’t shake the feeling that the AI industry learned half of its best moves from games. And honestly, teams can too. Somewhere between playtesting and product planning, between patch notes and process updates, the line blurs. That’s the thread I want to pull.
And if I can leave you with one nudge this morning: try Comet. It’s an agentic browser that feels like supervised self-driving for the web. You stay in control while it does the legwork, and you can watch it work. Notes below in Tool of the Week.
Shall we? 👇
🔦 How games powered today’s AI
Hardware built for play unlocked today’s AI. The same graphics chips that made games gorgeous and fast turned out to be perfect for the math behind modern AI.
Gaming demand paid for the muscle; AI now uses it at full tilt. In parallel, talent flowed from game worlds into AI, from researchers who taught AIs to play Atari to leaders now steering frontier-model programs.
And the “safe playgrounds” of game engines and simulators gave AI teams a place to learn by doing before touching the real world.
⚡ Quick Hits
🧠 Claude Memory launches: personalization you can manage. This is the “shared save file.” It keeps the useful stuff and lets you stay in control
📡 Talent signal: Microsoft AI is training its own frontier models with much larger GPU clusters in the works. Translation: the people who cut their teeth in game-like environments are now scaling the systems behind your copilots.
🕹️ At Gamescom, NVIDIA highlighted AI-driven NPCs and neural rendering moving from demos to dev kits. More proof that the game stack keeps seeding AI workflows.
🗞️ Fresh interview coverage has Microsoft’s AI Chief arguing for pragmatic boundaries (e.g., “machine consciousness” is a distraction). Useful framing for teams figuring out where human judgment stays central.
🛠 Tool
It feels like supervised self-driving for the web. You stay in the driver’s seat while the agent handles multi-step tasks and shows its work as it goes. Great for research sprints, compiling briefs, turning scattered links into a tidy summary with next steps and even reconciling your books.
😮 I’ve watched it complete an impressive array of tasks with dare I say a certain level of GRIT? Yes, that’s the word. See it for yourself!
🫵 Let’s share our coolests tests with each other. I’d love to see what you come up with!
🪨 The Pebble
This week, try one tiny playtest where people come first and technology makes it smoother between us. Maybe it is a hand-off, a planning thread, or a quick research sprint in Comet. One loop. One change. One win to share.
👂Where could you use that little nudge today so the rest of the week flows?
🎒 Takeaway
If the concert reminded me of anything, it’s this: nothing beats people in sync. Games gave AI the muscle and the playground, but it’s the humans who make it music. At Ampersand, that shows up in the way we plan, playtest, patch, and keep moving.
Thanks for reading all the way here. I appreciate you.
Before you hit the weekend pic, the poll, and go about your day, take a breath, send someone a quick thank-you, and then let’s go crush this week.
— Alejandro Jaegerman

😎 Looking Forward, to our next podcast episode that we need to record. Yes I’m using this opportunity to call this out Jen. So much going on! Let’s circle back to the mics!
😄 Let’s do it this week. What do you say?
📊 Poll
When was the last time you ran a "play" test? |
✉️ If someone forwarded this to you, you can subscribe at MondayThought.com.
[#19] 👋 Feeling momentum 🚀
Reply