External link for Nobodywho
Our CEO Cecilie Waagner Falkenstrøm is in San Francisco for TECH WEEK by a16z to discuss all things Nobodywho! Go say hello 👋
In San Francisco for Tech Week 🚀 If you’re around, let’s catch up. And ping me with any interesting events I should check out ☺️
Unsure where Nobodywho sits in tech stack and why you should consider it over Cloud based AI? Here is all you need to know in 40s form Ottilly Mould talking at our Mesh Community x Matrikel1 event. Headline is: no data leakage, quicker response time, no out of control costs. #aistartup #localai #cloudvslocal
We've had a little re-fresh 🍁 NobodyWho has had a bit of a makeover, so that our style matches our growing company. Let us know what you think! https://bit.ly/4dx8NVG
Europe is running on AI infrastructure it doesn’t own That was the premise of our event on EU Digital Resilience with Mesh Community collective last week at Matrikel1. Key points discussed: 💰There is a huge (€250b +) spending gap between the EU and US in terms of AI ☁️100% of EU's cloud is from US and Chinese providers ⚠️Cloud based AI open's us up to costs and data risks 🤖Local AI saves the day ;) Thank you to Sofie Käll and Christel Piron for being on the panel with us, Cecilie Waagner Falkenstrøm Ottilly Mould Asbjørn Olling for hosting/speaking. If you couldn't make it reach out to Ottilly Mould to hear more.
We've had a fun week at Nobodywho! ⭐ We went viral on The Hacker News (link below) 🍁 We launched a new branding/website (stay tuned) 🤖 We co-hosted (an over subsribed) event with Mesh Community at Matrikel1 👣 We worked from our investors The Footprint Firm as our office was out of action- thank you for having us! https://bit.ly/4hmXX5T
Nobodywho reposted this
Last Monday IDA Embedded ran our first "Get started running LLM local" hands on workshop. To take us through the jungle we had Asbjørn Olling and Marek Hradil to show the easy path using NobodyWho's Open Source library. They solve the problem embedding LLMs to your application, website fully no strings attached. We took a deep into how the open weight models are constructed and how local inference work and can be setup on you laptop or servers. For that part we had setup an isolated network with own firewall, server and some AI compute server. One was your sons gamer machine with full RGB ligth on. And NobodyWho had a brand new Nvidia DGX Spare connected to our network for every body to try models running on that one. Very positive feedback from the participants and request from some to dive even deeper. There are still a lot of things to explore. Since Monday I have updated my AI workstation with a used RTX 3090 TI, experimented with different models and optimized the token throughput with all the tricks Mads Henrichsen pulled of in Verbose Podcast #119. Tested Codex , Opencode and ended up with Pi and Gwen3.8-27b. And for my daily task it gets the job done. Sure my context window is smaller, but that just a reminder to work smarter. Split task in to smaller chuncks, build better skills. It just works will all my custom Claude Code skills that describe my IT setup, mail access, hardware lab equipment etc. The change over will not be hard ... Thanks to all who showed up - to Asbjørn and Marek for helping out and to the Verbose team Kasper Junge, Jonas Høgh Kyhse-Andersen for showing the way.
'As AI rises, the value of humanity rises with it' That was the heart of our co-founder Cecilie Waagner Falkenstrøm's talk at TechBBQ and forms the foundation of NobodyWho: the future of AI isn't only about scale, it's about critical reflection and the human decisions behind every system. While the rest of the world moved to the cloud, we became specialists in local. We believe that cloud doesnt have to be the default. Local, open, and human by design is the future we're building. #aistartup #localai #founderstory
Most automated tasks do not need the biggest model you can get your hands on. That has come up again and again in sales conversations lately: teams reach for the Ferrari when the task really calls for a bicycle. If the job is classification, extraction, routing, summarization, the best choice is often a smaller open-source models that is tuned for the work and not the latest Claude release. Why? - lower cost - lower latency - easier deployment - more predictable performance - a setup that is easier to run locally The real question is not “what is the most powerful model?” its what is the appropriate size model for this task? That is where local AI gets interesting because when you match the model to the job, you can keep things fast, private by design, and efficient without overbuilding from day one. Right-sized models win more often than people think. #LocalAI #Opensource
Want to hear more about the benefits of switching to local AI? Come meet us at our NobodyWho x Mesh Community event happening in Copenhagen in a few weeks. Cecilie Waagner Falkenstrøm will chat all things local AI, our investors Sofie Käll and Christel Piron will discuss why they invested in us/think local AI is the future. We will then wrap things up with a demo and a chance to meet the team and chat tech :). Request a spot in the comments below 🔗 #LLM #localai #aistartup
We are slowly coming down from the hype of TechBBQ and our main takeaway is: Europe really is such a strong market place for innovation. We met so many great start ups, EU investors and leaders who are flying the flag. Special thanks to The Footprint Firm & Norrsken Evolve for bringing together a equally inspiring and like-minded group to chat impact x startups. And to Christel Piron from PSV Tech for the fireside chat with Cecilie Waagner Falkenstrøm. Thank you to all that came and said hello and please do reach out to Ottilly Mould if you missed us and want to hear more about our mission at NobodyWho to keep AI sovereign, democratic and low cost 🤖.