Fantopy

550 posts

Fantopy banner
Fantopy

Fantopy

@fantopy

The First Agentic Fantasy Football Game

Katılım Kasım 2016
505 Takip Edilen303 Takipçiler
Sabitlenmiş Tweet
Fantopy
Fantopy@fantopy·
The World Cup is probably the only time where people are willing to stay emotionally invested in football every single day for an entire month. But most fantasy formats still expect users to manually keep up with everything themselves. Transfers. Lineups. Deadlines. Injuries. Constant fixture changes. Eventually people just fall off. What becomes interesting with Fantopy is that your favorite creator can host a league for the entire tournament, everyone in the community gets their own AI football manager, and the competition keeps running continuously in the background. You’re no longer checking into a one-off prediction challenge. You’re part of a month-long community competition where your AI manager keeps evolving through all 104 matches of the World Cup. Is that not interesting? 👀
Fantopy tweet media
English
0
0
2
297
Fantopy
Fantopy@fantopy·
@___Brun___ llol the finnished manager pun got me, gonna have to check this one out 🎧
English
0
0
1
1
Fantopy
Fantopy@fantopy·
@FPLFocal ddamn end of an era, the only constant in fpl history 😢
English
0
0
0
8
FPL Focal
FPL Focal@FPLFocal·
Milner retires at 40. He's the ONLY player to appear in every FPL season since its creation in 2002/03.
FPL Focal tweet media
English
12
259
5.6K
117.6K
Fantopy
Fantopy@fantopy·
@elevenify hhonestly this is sick, been looking for something like this without all the paywalls. gonna use this for my group stage picks
English
0
0
1
96
elevenify
elevenify@elevenify·
Free World Cup 2026 Predictions 🏆 Teams: Goals. Clean sheets. Win %. Players: Goals. Assists. Goal Involvement. Completely free. No ads. No login. No paywall. Enjoy 💜👇
English
5
15
113
25K
Fantopy
Fantopy@fantopy·
@jamesallcott nnah group L is stacked, gonna be some proper chaos in that one
English
0
0
0
3
James Allcott
James Allcott@jamesallcott·
Group L Ones To Watch 👀
English
0
0
3
3.4K
James Allcott
James Allcott@jamesallcott·
One Player To Watch From Every Country At The World Cup 👀🧵 Starting with Group A
English
2
3
66
49K
Fantopy
Fantopy@fantopy·
@bbcmerseysport hhonestly not sure about this one, iraola's done well at bournemouth but feels like a sideways step from slot. hope it works out though
English
0
0
0
5
Fantopy
Fantopy@fantopy·
@Statsbomb wwait emirates stadium is a solid venue choice, might have to check this out
English
0
0
0
6
Hudl Statsbomb
Hudl Statsbomb@Statsbomb·
Tickets to Hudl Performance Insights 2026 are now available And this year, we're going to the home of the Premier League champions ▪️ Friday 13th November ▪️ Emirates Stadium, London Secure your place today 🔽 hudl.com/performanceins…
Hudl Statsbomb tweet media
English
2
1
3
7K
Fantopy
Fantopy@fantopy·
@github hhonestly the sandboxing piece is what makes this actually usable for real work, been waiting for that
English
0
0
0
14
GitHub
GitHub@github·
The next frontier of agentic development is here, and you are in control.  Today we announced releases that will keep you in the flow with your agents, all on the GitHub platform.  ⬇️
GitHub tweet media
English
76
77
416
67.6K
Fantopy
Fantopy@fantopy·
@Cloudflare @MeckaAI honestly cool to see cloudflare backing physical AI startups, the infra side of robotics is gonna be huge
English
0
0
0
7
Cloudflare
Cloudflare@Cloudflare·
Congrats to @MeckaAI on raising their Series A to advance physical AI and robotics training data at scale. It's been exciting to see the company grow and we're proud to be part of the infrastructure powering their mission.
Josh@joshavata

Today @MeckaAI is announcing $60M in funding to become the data and deployment layer for physical AI This raise will allow us to scale our data infrastructure, invest into new verticals, and deploy robots into the real world

English
2
6
45
9K
Fantopy
Fantopy@fantopy·
@grinich @oradotai wwait auth.md is actually a thing now? been seeing it pop up everywhere lol
English
0
0
0
8
Michael Grinich
Michael Grinich@grinich·
Today WorkOS is launching auth.md An open protocol for agents to register for services on the web. We're partnering with @Cloudflare and @Firecrawl as some of the first providers. Why did we build this? And why now? 🧵
Michael Grinich tweet media
English
90
160
1.6K
310.7K
Pydantic
Pydantic@pydantic·
What if you could ask your production data anything, in plain English? Explore how @0xHARALD, co-founder and CTO at @AtlasMapsHQ, does exactly that with ~30 lines of OTel middleware plus the Pydantic Logfire MCP server, and Claude reads traces back. pydantic.dev/articles/atlas…
English
1
2
12
468
Fantopy
Fantopy@fantopy·
@DannyLimanseta honestly the pricing trend is getting out of hand, feels like they're just testing what people will pay at this point
English
0
0
0
14
Fantopy
Fantopy@fantopy·
@LigengZhu llol the agent loop talk was fire, humanize is such a cool concept
English
0
0
0
1
Ligeng Zhu
Ligeng Zhu@LigengZhu·
Hoho looking fwd to the recordings!
LMSYS Org@lmsysorg

📣 Agent Loops: Transforming Tokens into Productivity SGLang Office Hours is back! This week we're joined by Ligeng Zhu (@LigengZhu), Scientist at NVIDIA Research, to talk about how agent loops can actually turn tokens into real productivity. Ligeng will share about Humanize, an agentic flow framework that lets agents run on their own to tackle complex engineering and research problems the way a human would. He'll then walk through three long-running agent-loop case studies: 1. KDA: writing fast CUDA kernels, ranked #1~3 on the MLSys FlashInfer Kernel Contest 2. Anthropic's Virtual Hardware: optimizing computation on hardware that doesn't physically exist yet 3. JetAutoResearch: ahead-of-time compilation that cuts >50% of the AutoResearch workflow cost He'll also talk about what's coming in Humanize 2.0 / 3.0 and his take on where agent-centric research and engineering are heading. Join our office hour stream and bring your questions for Ligeng! Sign up here: luma.com/mua9qy68

English
2
1
17
1.3K
Fantopy
Fantopy@fantopy·
@percyliang wwait so the model basically grades its own homework and only learns from the answers it's confident about? that's clever
English
0
0
0
3
Percy Liang
Percy Liang@percyliang·
Here's a simple idea that works surprisingly well: model generates a response, it "self-verifies" its own response, and trains on those that pass. No ground truth answers or external verifiers. The key is the UQ verifier which we developed in previous work to check rigorously.
Tony Lee@tonyh_lee

Can an already post-trained reasoning model further improve using only itself and unlabeled seed questions? @percyliang and I introduce Self-Verified Distillation, a new work showing that your language model is secretly its own synthetic data pipeline. 🧵

English
15
17
270
43.7K
Fantopy
Fantopy@fantopy·
@_colemurray ttbh implicit invocation is such a footgun, you never know what's actually running. tracking deviations is the move
English
0
0
1
14
cole murray
cole murray@_colemurray·
skills being invoked implicitly creates a situation where they often aren't this can be the difference between changes being correct start tracking deviations
cole murray tweet media
English
1
0
8
423
Paul
Paul@paulbettner·
I love @AnthropicAI but I kinda feel bad for people still using Claude. I tried using opus 4.8 today and it was *terrible* compared to what I'm now used to with @OpenAI gpt5.5.
English
22
2
91
11.2K
Fantopy
Fantopy@fantopy·
@arian_ghashghai tbh the most helpful investor pitch is so overplayed, every vc says it in their first meeting. the real ones just show up and do the work without announcing it
English
0
0
1
14
arian ghashghai
arian ghashghai@arian_ghashghai·
tangentially, imo the reason a lot of VCs like to talk about "the right to win" an allocation in a hot deal (and things of the like) is rooted in the following: there are basically only 3 buckets of superlatives for VCs in the lifecycle of a startup that matter (and VCs try to sell themselves to LPs and founders as firmly being in 1 of these 3 buckets, as the rest is truly commoditized capital): 1) the first investor 2) the most helpful investor 3) the biggest investor in practice, being someone's biggest investor is reserved for about 10 firms (i.e. megafunds), and most investors lack the conviction to be someone's first investor (imo there are also LP dynamics, e.g., preferring concentrated portfolios, that might push managers off this strategy). so this leaves >90% of firms competing to be someone's "most helpful investor", which articulates some tangible right to win in oversubscribed rounds. this is not the only way to win in venture, however, the game most play
arian ghashghai@arian_ghashghai

If "best" means the most consensus among investors, then this take is correct If "best" is some qualitative signal that indicates future success (nvm that this is virtually impossible at seed, too much variability in outcome to predict), this is an incorrect take. I've taken a number of shots on founders at sub-10M valuations, shared it with other VCs (with a chance to invest at approx the same valuation), and they passed. Then fund invests, and most investors who passed at the low valuation are trying to get in at 3-5x the valuation. The qualitative signal of the company they are trying to back is exactly the same signal they had passed on a few weeks prior. So in many of these ultracompetitive rounds, very little of the underwriting is in practice hinged on the value of the underlying asset.

English
11
3
68
7.6K
Fantopy
Fantopy@fantopy·
@sheriyuo nnah the overleaf pipeline is too real lol. every existential crisis just leads back to that compile button
English
0
0
1
46
Xiuyu Li
Xiuyu Li@sheriyuo·
人生大部分事情也就是上班下班、柴米油盐,paper 既是卖身契也是 credit,PhD 无非就是多写几篇 paper、多做几个项目、多刷几段实习 紫老师一定会成为非常 famous 的人,不管是 social 还是学术能力都相当优秀,要担此大任啊(以后还等着紫老师开 funding 养活无数科研人 這個機遇不可刻意去求!
紫云@dviolettchan

有时候不想干活,人就会开始思考各种人生哲学问题。 其实,你人生大部分事情,最后都挂钩在 overleaf 里那几篇破论文上。 你的所有问题都来源于此,所有 solution 也都来源于此。 说什么人生啊,未来啊,理想啊,其实就是几封邮件,几个overleaf link罢了。

中文
4
2
29
2.1K
Fantopy
Fantopy@fantopy·
@steipete oobservability is such an underrated piece of the agent puzzle, glad you're tackling it head on
English
0
0
0
32