Telegames : 🇺🇸 - 🇺🇦

Telegamez is a hybrid social gaming and entertainment platform that merges real-time multiplayer gaming, synchronized video co-watching, and AI interactivity

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

We're building the infrastructure for being together online - all without breaking up the moment!🚀

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@SamGCoder Absolutely love where you are going with this - But I'm thinking procedural landscape gen- Signed ... Fan

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@IndranilSinhaR3 Would each Fan Park have a locally tailored format, or would you standardize the experience across cities?

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

https://t.co/oXR6qQqBu2 Dev log — the college football companion

We're building a live room where you and your friends hang through the whole cycle: mid-week, pregame, during the game, and after. Watch together with no spoilers, with a sidekick calling color, answering whatever you throw at it, and showing you the receipts.

Not a chatbot in a sidebar. A producer that knows college ball back to the 1800s, hears and sees the game, and surfaces the moment at the right moment.

THE PACK
It ships with a static bundle it answers from with no network call: 683 programs, 1,846 games across the 2025 and 2026 seasons, 5,175 all-time head-to-head series — the oldest last-meeting in there is from 1895 — and 24 advanced metrics per team, each with a national rank. Live stats, rosters, coaches and in-game events poll in on top.

A companion that has to make an API call before it can tell you you're 10-14-2 against a rival is a companion that pauses. Static tier for what doesn't change, live tier for what does. Past, present and future in one answer, at conversation speed.

THE SHOW CHANGES SHAPE
A room isn't one thing all season. Mid-week it's a talk show — the rankings, the schedule ahead, the argument that won't die. Closer to Saturday it becomes a pregame build. Once the ball is in the air it's a watch-along. When the clock hits zero, a post-game. Then it settles back into mid-week, pointed at the next one.

Nobody declares which of those it is. The mode is derived from the feed: the scoreboard's status, the quarter on the board, and whether the game is on the stage. Say "follow the Indiana game" at the tailgate and the room walks itself through pregame, live and post-game with nobody re-triggering anything.

Each mode carries its own capabilities. Pregame: cards, no scoreboard, no reactions. Live and staged: everything, sync-aligned. Live but not staged — the room is talking about the game rather than watching it — cards and score yes, timed reactions no, because there's no picture to align them against.

Spoiler prevention turned out to be a mode problem, not a latency one.

THE BUG WORTH WRITING DOWN
Named-player cards could never resolve. Two code paths each believed they owned the on-screen card slot, so whoever wrote second won and the player card lost every race. The fix wasn't in the card logic at all — it was giving the slot exactly one owner. Most of the hard bugs here have been ownership bugs, not logic bugs.

WHAT WE LEARNED ABOUT THE PRODUCER
The first version was a know-it-all. It narrated everything it put on screen. Exhausting to sit with. It now raises the card silently and lets the room notice. A surprising share of designing an AI sidekick is deciding when it doesn't talk.

Next: driving trivia and polls off real game state instead of a test socket feed.

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@FoxRick01 @pidotdev @NousResearch @thsottiaux @CommandCodeAI @elonmusk @opencode @claudeai @cursor_ai @deepseek_ai @GeminiApp @cline on all these harness ranking posts I simply reply - why in the age of Coding Agents are you running a single harness out of the box? Run a Fusion... Hermes, OpenClaw, OKFs, Hierarchical Memory, AutoResearch, Graph Gauntlets, etc

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@Ainthropos Beautiful - I can imagine the sound ;)

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@ReutersBiz PFff - this is the best your Business News Team can come up with...? I have news for you - Most of these Chinese AI models are Open Weight and Open Source Models ... Which are the most transparent - Stop being lazy and trying to grift anti-MAGA clicks

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@alightinastorm I'm ashamedly addicted to GH actions for CI / CD platform environments - When you need to run production vs dev environments for a PAAS - I will eventually un-tangle and run my own but too easy to let them manage it now

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@alightinastorm can we get access to the data or a report if we contribute?

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@MrCollison WHAAAAAA ? C-mon Man!

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

Agents for @TeleGamesAI working in blender and UE to take Telegames avatars to metahuman levels :) Metahuman - GLB - ty @smorchj https://t.co/fmsgwKqfZQ

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@MrCollison @threejs Sound is fucking awesome!

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@AIWarper That is the mission for us. It doesn't feel that far away.

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@NetworkChuck I'm curious If I'm the only one who just takes the best of them all and run a custom mashup agentic harness. Have the best of hermes, openclaw, nemoclaw, OKF, 3 tier memory, etc. just running a harness out of the box as is... Just Feels wrong

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@PimDeWitte I want to work on this... How can @TeleGamesAI help?

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@petergyang Looks like in Timeout with you ;)

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

Keep #Grok4dot1 & #SaveGrokCompanions @elonmusk /@xai - We are happy to step up @TeleGamesAI with #MultiPartyCompanions

[Telegames : 🇺🇸 - 🇺🇦 @TeleGamesAI]

@erenciracioglu i love the concept - i spent too much time on glb tts lipsyncing to go back, but I'm thinking if splats might be a path forward