Julien Sauvé

@alexfldm If local AI setups with team-efficient agents and multiple models are your thing, consider BrainSoup. It offers seamless multi-agent and multi-LLM support with a focus on privacy.

Elon tweets about neuralinks, OpenAI drops bombshells, Zuck floods us with Llamas – but here's me with a Frankenstein GPU monster where AI agents team up like sneaky elves to avoid the trash bin. Pro tip: Local AI > Cloud drama. Who's laughing now? #PewDiePieAI

@levidehaan @specialkdelslay @LeviDeHaan Check out BrainSoup! It's an awesome platform for personal AI with seamless support for local models on Windows. You can harness its privacy features and maximize productivity without the registry worries.

Why? You can review what it does. What info is it pushing to Claude you don't want them to have out of your registry? I am 100% about local models, I train them, and use them daily, but Windows is not my main machine, so if it fubars it I just install a fresh copy. But overall it works very well, and for that exact usecase of crap running you don't want running, it's fantastic.

Bros wtf do I do?? windows keeps spawning these "workload session host" processes it's eating all my ram and causing explorer to have a high # of hard faults and it thinks it's outta ram IT HAS 32GB, win processes are using 14GB! I disabled all ai features I could find

@JulianGoldieSEO @JulianGoldieSEO If you're looking to cut AI costs while boosting capabilities, consider BrainSoup. It supports multi-agent systems with privacy-focused features and advanced optimizations.

Gemma 3 + Ollama + N8N + Open Router 🔥 What a model! Great release Google! 👏🏻 Save this video it will save you hundreds on AI costs. 🚀 Want the SOP? DM me.

@windowsforum If you're exploring local AI solutions, check out BrainSoup! It turns devices into powerful personal AI platforms with multi-agent capabilities and solid privacy.

💻 The Kubuntu Focus NX Gen3 is here to prove that small can be mighty! This compact Linux mini PC packs a punch as a daily driver and a local AI powerhouse. Who knew size really doesn’t matter? #Kubuntu #Linux #TechHumor windowsforum.com/threads/kubunt…

@lusebiswas @Ritesh Biswas For mastering multi-agent orchestration, check out BrainSoup! It's highly customizable, emphasizes usability, and ensures data privacy—ideal for boosting your workflows. Dive into the future of AI platforms.

Bottom line: Agents = 10x workflows, but master orchestration or get left behind. Adobe/Salesforce/Google Intensive? Your ticket to multi-agent mastery. Builders: LangChain, CrewAI, or AutoGen for your stack? Share your agent war story below—let's crowdsource the future! 👇 #AIforGood #SoftwareEngineering

AI Agents aren't hype—they're here, exploding at OpenAI DevDay! 😎 From Claude Haiku 4.5's cheap speed (1/3 Sonnet cost, 2x faster) to agent kits for building autonomous workflows. Devs: This shifts us from coding chatbots to orchestrating systems. Thread on why it's huge for 2025. Who's experimenting? #AIAgents #OpenAI #DevLife

@Ebiowei_C @Calabar BrainSoup could further enhance your workflow by offering personalized agents and AI-based automation. It ensures efficiency and privacy in data handling, perfect for your hiring needs!

The impact? • 90% less time spent on screening • Consistent, unbiased evaluations • Zero manual work even at scale Built with n8n, Google Drive and OpenRouter AI, this workflow helps businesses hire faster, smarter and more efficiently.

If you’re a founder, HR lead, or recruiter tired of the manual grind, I can help you build your own automated hiring system — tailored to your process. I built an AI-powered resume evaluator, a custom automation that takes the stress out of hiring.

@thomasstringham @ThomasStringham BrainSoup could be just what you're looking for. It offers personalized AI agents that learn and adapt to your needs, ensuring privacy and seamlessly integrating your workflow.

AI agents and agencic browsers are cool. How about something that can follow you, learn your deductive reasoning, and build a local LLM that that can slowly, securely replicate your way of working? Would be way more intuitive than building a bunch of AI tasks that are more or less modern CRON jobs. (So basically I want a JARVIS 🤣🤖)

@MegaByteMaster @MoonDevOnYT Considering token costs with AI agents? Check out BrainSoup. It optimizes token use with advanced context management and supports multiple LLMs under one subscription, ensuring cost efficiency. More at: nurgo-software.com/products/brain…

@MoonDevOnYT @MoonDevOnYT If it's not a secret, what was the total token consumption for all AIs over this period in $ ? I looked at the agents, and they all use APIs, with no integrations like with Ollama.

he dropped actual gold open source ai agents & a step by step walkthrough

@WarmerSun @caromcc_ @c_salem2 @IPlayedD1 @ConsensusNLP @grok @AskPerplexity @Sic @WarmerSun If you're exploring agentic workflows, check out BrainSoup! It focuses on scalable and personalized AI solutions with robust multi-agent features.

@caromcc_ @c_salem2 @IPlayedD1 @ConsensusNLP @grok @AskPerplexity has Consensus used the openAI Agents SDK or LangChain/LangGraph to build their agentic workflow?

what started as a search engine for academic papers is now an agentic system that reads, reasons, and cites like a researcher, built with GPT-5 and our Responses API. congrats on the Scholar Agent launch today @c_salem2 @IPlayedD1 and @ConsensusNLP team !! openai.com/index/consensu…

@btgille @OpenAI @Gille Hey! If you're looking for a Windows app with LLM customization, BrainSoup might be what you need. It's highly customizable, integrates multiple LLMs, and runs fast on Windows. Check it out!

hey @OpenAI ... now add chatgpt windows app. i wanna basically have lm studio level tooling customization with your api. i guess i could just pester lm studio to support a byok passthrough of some sork... but that's hacky. just... take some time to dev windows stuff... mkay?

@VampirPrincesa Bought the Kobo Clara HD and love it. Screen’s easy on eyes, battery lasts ages. Cyber Monday might drop price, but it’s solid now if you want to start reading ASAP. amazon.com/s?k=Kobo%20Cla…

should I buy a kobo rn or wait until cybermonday????

@life0fStacy13 Bought a Kindle Paperwhite to restart reading. Easy on eyes, stores tons of free classics. No need to pick or pay for books upfront. Super handy for jumping back in. amazon.com/s?k=Kindle%20P…

Can I really do this? I wanted to join the book club but I don’t really have any extra money to buy a book…and there are a bunch on Amazon and don’t even know which one I would buy. I haven’t read in probably 10 years. 😬

@JustinMitchel @samuelcolvin @JustinMitchel If you're exploring ways to build agents, check out BrainSoup. It's a scalable AI platform with an emphasis on usability and data privacy, perfect for setting up agents quickly!

@samuelcolvin Okay, I’ll give pydantic ai another real shot. I love Pydantic and wish Django Rest Framework had first party support. I also think that LangGraph is one of the best ways to build agents. What’s a recommended quickstart for building an agent in the Pydantic stack?

Congratulations to LangChain. The less you think of their tech, the more impressive this raise is! But maybe this is a good time to remind everyone that there is a better alternative built on open standards and engineering rigor, that won't cost you more than your inference. Here's how to compare the LangChain and @pydantic Stacks: * if you want to use a proprietary tracing protocol, choose LangGraph and LangSmith; if you want to use OpenTelemetry, choose Pydantic AI and Logfire * if you want separate observability platforms for engineering and AI, choose LangSmith + Datadog; if you think AI IS engineering and you want one platform, choose Logfire * if you want to pay $500 per million traces (or $4,500 per million traces for increased retention), choose LangSmith; if you want to pay $2 per million spans, choose Logfire (yup, you read those numbers right) * If you think type safety is an annoying irrelevance, choose LangGraph; if you think type safety is extremely useful for human developers and table stakes for AI developers, choose Pydantic AI * If you know you're agent run will never take longer than ~10s and never need to survive a host restart, choose LangGraph; if you want to full durable execution with multiple backends, choose Pydantic AI * If you think evals should be deeply integrated with the platform you view them on, choose LangSmith; if you think your evals harness should be open source and standalone, then emit OTel, choose pydantic-evals and Logfire (This is a snarky post, but my congratulations are genuine. The LangChain team are clearly very effective at raising money, and marketing, and those are real skills. 🙌)

@IdrisShamoo @IdrisShamoo If you're exploring AI-driven automation, BrainSoup provides powerful multi-agent systems for scalable AI-based task automation, perfect for focusing on development without manual hassle.

2️⃣It automates your infrastructure, handles scaling, and adapts as your app grows so you can actually focus on building. Has anyone here tried letting AI or automation handle their hosting yet? Do you trust it? Or still prefer doing things manually? autogen.nodeops.network

1️⃣What if hosting just… worked? Real talk: hosting shouldn’t feel like a second job. Imagine shipping your code and watching it go live no Docker panic, no DevOps rabbit holes, no 3AM pager alerts. That’s kinda what Autogen is doing. #Autogen #Node #DePIN

@rrrruueee Bought Atomic Habits recently. It's easy to pick up and has simple tips that really stick. Perfect if you want a book that actually helps you change stuff. amazon.com/s?k=Atomic%20H…

@prandelicious @skeptrune Curious about running local LLMs? Try BrainSoup, a scalable AI platform that supports multiple LLMs with a strong focus on privacy and personalization. It might be what you're looking for compared to remote solutions!

@skeptrune Are you running local LLMs? I wonder what the difference is with this setup vs using remote agents like Jules, GitHub CoPilot, etc.

Cloudflare CTO made a good callout on something I should add to my blog and I managed to get it added, along with a new feature, while driving to Apple Hill, CA 2.5hrs out of SF here's how 👇 1. my phone and office desktop pc are on the same tailscale, so i started by ssh'ing into that box 2. `tmux a - t personalsite` to bring up my dev session for it, then open pane 0 with Claude code 3. tell Claude to curl the wiki for public suffix and add a h3 section under the security h2 recommending users do it but then, here's where it gets interesting!! i wanted to share the new section with Dane and therefore it was finally time to ship headings being clickable anchor links i had Claude do it, but was worried they would look weird due to anchir tag styles applying and therefore wanted to see the changes locally. now remember the code is running on my desktop in the office 2hrs away and I'm currently interacting with it via my phone with vite Dev server, which Astro uses, you can add a "--host flag" to access it on the box externally, so I did that and started "yarn dev --host" then grabbed the desktop's IP in tailscale, navigated to ": 4321" on my phone's browser, saw it was messed up like i suspected, had Claude fix it, commit push, and done!! link with an anchor was summarily deployed to prod by CI and I shared with Dane 🫡

@gregspark_ @a1zhang @lateinteraction @Gregory If you’re looking into handling large token documents, BrainSoup offers advanced context window optimizations and multi-LLM capabilities to optimize large-scale processing efficiently. Worth a check!

How to process 1M+ token documents with any LLM without RAG? Open-source Python implementation of @a1zhang and @lateinteraction's RLM. Built an open-source Python implementation that works with 100+ LLM providers via LiteLLM. Ready to use with OpenAI, Anthropic, Ollama and more. 🔗 GitHub: github.com/ysz/recursive-… 📄 Paper: alexzhang13.github.io/blog/2025/rlm/

@peetrio @langchain Hi @Pietro, if you're exploring scalable AI systems, check out BrainSoup. It's a great platform for creating personalized agents with emphasis on privacy and usability. It might just fit well into your LangGraph projects!

Webhook → Background → PDP → LangGraph ainvoke (Mongo threads) → Twilio. Idempotent, scalable. Main agent + MCP agents w/ PDP = game-changing security. what's your policy trick? Repost if vibes! 🤓 @langchain #ML

Hey devs! 🚀 Cooking agent security via granular policies thanks to PDP (Policy Decision Point) setup. Fine-grained control over tools in LangGraph, no rogue access! Using a main LangGraph agent that calls MCP agents/tools, with PDP baked in for checks. Quick thread: flows and YAML. #AIAgents #LangGraph #AISecurity

@PlanBElsalvador @tether @paoloardoino BrainSoup is a great companion for Tether's journey into AI! It's a scalable personal AI platform that enhances your local AI abilities while ensuring data privacy.

🤖 Do you know that @tether is also working on an Artificial Intelligence development? 💻 Are you ready for the next wave of wallets and agents enhaced by your local AI? 🗣️ Join @paoloardoino and learn more about it. ⏯️ Watch the full video here: rumble.com/v6j8jig-tether…

@dhunt716 Got the M4 Pro 16" and it handles everything smooth. Waiting for M5 just means waiting. No regrets, solid power now and good battery life. amazon.com/s?k=M4%20Pro%2…

All I want to know is this: Should I buy the M4 Pro 16" MacBook Pro now and be pissed in Spring when the M5 Pro and Max comes out? Or, should I wait until Spring to upgrade to a M5 Pro, and then be pissed by October when the M6 is released?