Replying to @ycombinator
@ycombinator's @dessaigne is joining us at Flower AI Day 2026.
Nicolas is a General Partner at YC and co-founder of Algolia, the search API company that went through YC’s Winter 2014 batch. As CEO, he grew @algolia to a team of 350 before handing over the role in 2020. Today, he works with founders at Y Combinator, bringing experience from both building a developer-focused company and helping others build theirs. His background also includes a PhD in Computer Science from the @NantesUniv.
Join Nicolas and the researchers, engineers, and founders at Flower AI Day on September 30 at The Pearl in San Francisco. We’ll spend the afternoon and evening exploring open, collaborative AI through technical talks, demos, firesides, and conversations with the people building it.
Register: luma.com/flwrlabs-uygg
I am happy to share a new paper.
Flower Hub is an important step toward an “App Store moment” for Collaborative AI: A Reproducible Benchmarking Platform for Federated Learning in Simulation and Deployment.
A huge thank you to all our co-authors: @jafermarq, @dstripelis_ , @lorenzosani97, Davide Eynard, Fan Zhang, @HongJia_1 , Ting Dang, D. B. Emerson, Fatemeh Tavakoli, Ole Werger, Lars Wulfert, Petros Demetrakopoulos, Sofia Tsekeridou, InSeo Song, KangYoon Lee, Honghao Li
More apps and agents are being published, and the direction is clear: away from isolated experiments and toward an open, reusable ecosystem for Collaborative AI.
1
11
By giving federated systems, decentralized applications, and collaborative agents a common way to be packaged and shared, Flower Hub allows researchers and developers to build on one another’s work instead of repeatedly starting from scratch.
14
Flower Hub started earlier this year as a place to share federated apps. It has grown into a Hub with 137 applications and 23 agents.
22
A big milestone for Flower! Endeavor 1.0 brings frontier-class reasoning, coding, and agent capabilities together with something enterprises increasingly need: control over how and where their AI runs.
Frontier AI shouldn’t have to mean permanent dependency on a closed API.
Introducing Endeavor 1.0 by Flower.
Frontier AI, under your control.
Our most powerful model yet. Frontier-class reasoning, coding and long-horizon agent capability, competitive with leading models from OpenAI and Anthropic.
Use Endeavor through Flower, or deploy it on infrastructure you control.
92.0 GPQA · 98.2 HumanEval · 99.9 AIME 2026 · 94.1 IFEval
This is Endeavor.
Enterprise-ready agents need enterprise-grade evals: real workflows, private tools, domain rules, and measurable outcomes without centralizing sensitive traces.
That’s FlowerBench.
Explore the leaderboard and submit your task.
AI agents are advancing quickly.
But enterprise work remains a much harder test.
Today coinciding with ICML, we are introducing FlowerBench: a frontier agent benchmark for evaluating AI agents on secure, proprietary, long-horizon enterprise tasks.
FlowerBench evaluates agents where enterprise work actually happens: inside organizations, with private data, internal tools, domain rules, and clear success criteria.
Evaluations are coordinated through the Flower Enterprise Evaluation Network. Sensitive data and context stay within each organization. Only sanitized, non-sensitive results are shared.
Developed with opt-in early partners, FlowerBench makes it possible to benchmark agents on realistic enterprise workflows without centralizing sensitive private traces.
We have already assessed proprietary and open agents and models across a range of enterprise-grade tasks in various industries: Healthcare, Insurance, Operations, MLOps, Legal, Marketing and Finance.
FlowerBench introduces a new way to build enterprise agent benchmarks: privacy-preserving workflow collection, distributed evaluation, and an evaluation network grounded in real enterprise work.
We are inviting enterprises to contribute tasks and help shape the next generation of enterprise-ready AI agents.
Links to the launch blog post, how to contribute and the leaderboard itself in the thread below.
2
101
Proud to see Flower Hub launch today!
A hub for publishing, discovering, and running federated AI apps across environments — making decentralized AI easier to build, share, and deploy.
A big step toward a true Collaborative AI ecosystem.
Today, together with our launch partners, we’re proud to introduce Flower Hub -- an app hub for publishing, discovering, and running Flower apps across heterogeneous environments.
Flower Hub enables developers to focus on what matters most: building decentralized and federated AI applications. The underlying infrastructure -- from distribution to execution, across both simulation and real-world deployments -- is handled by Flower.
This launch represents a major step forward for collaborative AI. It marks the transition from a world of isolated projects to an open platform where a community can share, reuse, and build trusted apps together.
The era of collaborative AI starts now.
Flower Hub Launch Partners: Mozilla AI (@MozillaAI), Owkin (@Owkin_AI), Sony AI (@SonyAI_global), BloodCounts! (@blood_counts), Fraunhofer IMS (@FraunhoferIMS), Gachon University, SYNTHEMA (@SYNTHEMA_EU), the University of Cambridge, the University of Melbourne, and the Vector Institute (@VectorInstitute)
Today, together with our launch partners, we’re proud to introduce Flower Hub -- an app hub for publishing, discovering, and running Flower apps across heterogeneous environments.
Flower Hub enables developers to focus on what matters most: building decentralized and federated AI applications. The underlying infrastructure -- from distribution to execution, across both simulation and real-world deployments -- is handled by Flower.
This launch represents a major step forward for collaborative AI. It marks the transition from a world of isolated projects to an open platform where a community can share, reuse, and build trusted apps together.
The era of collaborative AI starts now.
Flower Hub Launch Partners: Mozilla AI (@MozillaAI), Owkin (@Owkin_AI), Sony AI (@SonyAI_global), BloodCounts! (@blood_counts), Fraunhofer IMS (@FraunhoferIMS), Gachon University, SYNTHEMA (@SYNTHEMA_EU), the University of Cambridge, the University of Melbourne, and the Vector Institute (@VectorInstitute)
I just arrived in San Diego for #NeurIPS2025, where I’ll be presenting our FlowerTune poster!
📅 Poster Session: 4 Dec
⏰ 4:30–7:30 PM
📍 Exhibit Hall C, D, E — Poster #800
Come stop by if you’re interested in discussing federated LLM fine-tuning for real-world applications!
📄 Read the paper: arxiv.org/abs/2506.02961
🏆 Explore the leaderboard: flower.ai/benchmarks/llm-lea…
1
38
We’re still open to new submissions and eager to collaborate on upcoming challenges, tasks, and benchmarks. If you have ideas or want to get involved, let’s chat—we’d love to work together to push federated LLM fine-tuning to the next level. 🚀
1
45
🌟 Exciting news! Our paper “FlowerTune: A Cross-Domain Benchmark for Federated Fine-Tuning of LLMs” has been accepted at NeurIPS 2025! Huge thanks to all co-authors for their incredible effort.
FlowerTune is the first comprehensive benchmark suite and LLM leaderboard for evaluating federated fine-tuning across four diverse domains: General NLP, Finance, Medical, and Code.
📄 Read the pre-print: arxiv.org/abs/2506.02961
🌐 Explore the leaderboard: flower.ai/benchmarks/llm-lea…
This week, I’ll be in San Francisco for:
1️⃣ Flower AI Day – Sep 25: flower.ai/events/flower-ai-d…
2️⃣ Decentralized AI Hackathon at Stanford University – Sep 26: flower.ai/events/decentraliz…
Come join us to learn more about FlowerTune and discuss federated fine-tuning of LLMs!
🚀 New Paper Alert: FlowerTune: A Cross-Domain Benchmark for Federated Fine-Tuning of LLMs — from @flwrlabs
We're thrilled to introduce FlowerTune, the first benchmark that evaluates federated fine-tuning of 26+ LLMs across 4 domains: NLP, Finance, Medical, & Code.
Co-authors (3/3): Daniel J. Beutel @daniel_janes, Lingjuan LYU @lingjuanlyu, Nicholas Lane @niclane7, and contributors from the community who made this possible!
3
129
Coauthors (2/3): Jiamu Bai, Daoyuan Chen, Zikai Zhang, Rui Hu, Inseo Song, KangYoon Lee, Hong Jia @HongJia_1 , Ting Dang @tingdang513, Junyan Wang @wjydd666, Zheyuan Liu
3
261
Huge thanks to all co-authors (1/3): Massimo Roberto Scamarcia, Javier Fernandez @jafermarq, Mohammad Naseri @mnaseri2770, Chong Shen Ng @chongshenng, Dimitris Stripelis @dstripelis_, Zexi Li @TomLeeeeee, Tao Shen
📄 Read the paper: arxiv.org/abs/2506.02961
🌐 Explore the leaderboard: flower.ai/benchmarks/llm-lea…
🔍 More research from Flower Labs: flower.ai/research
1
37











