Knut O. Hellan (@KnutHellan) header
Knut O. Hellan (@KnutHellan) profile picture

Co-founder of prop tech startup https://nitter.cf/t.co/niZ6gAK6Lu, interested in AI/ML, NLP/search, data strategy, scalability, skiing and cycling.

Oslo, Norway
Joined March 2008
Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)Photo by Knut O. Hellan (@KnutHellan)
Knut O. Hellan retweeted
Five years ago @adamvpro and I started @zenml_io because getting AI from a notebook into production was painful. It still is. Only now it's agents, not just ML models, that make it hard. Building an agent takes an afternoon. Keeping it working is the hard part. A prompt tweak breaks something that worked yesterday. A cheaper model quietly stops answering. You only hear about it when customers complain. Every team we talked to said the same thing: we don't need another dashboard, we need someone to do the work. So we built that. ZenML Pro now comes with an always-on AI engineering team. It reads your traces, finds what your evals miss, and proves the fix works before you ship it. This tech is backed by ZenML and Kitaru, our two OSS products, that stay open source and freely available. To kick off this new chapter, we have decided to rebrand. ZenML is now ZenML Labs, and our ambition is greater than ever. Connect your agent traces today, free at zenml.io. We are very proud of the onboarding ;-) We wrote about it here: zenml.io/blog/introducing-ze…
18
13
5
69
16,971
Knut O. Hellan retweeted
Revolut: Europe's most valuable startup ($200B) just showed the world its AI strategy. Every company will end up copying it. At the start of this year, Revolut leaned on the frontier labs. Its 2,000 engineers used Claude Code and Codex, and some burned $1,000 to $10,000 a month each. The bills kept climbing, and the returns weren't there. So Revolut switched its default to an open-source model called GLM. All it needs is GPUs, which it rents from Together AI. The results, by Revolut's own count: - The bill is about a tenth of what it paid OpenAI and Anthropic - Engineers get unlimited tokens instead of spending caps - Coding time is down 60% and lead time is down 70% That's the chart below. Revolut still uses Claude and GPT, but only as "the exception," for the tasks they do best. Everything else runs on open source. So here's the question I keep coming back to: how many use cases can companies really justify sending to frontier models? My answer is very few. Plenty of companies still route everything to OpenAI and Anthropic today. For how long? I think open source ends up handling the vast majority of corporate AI work, and frontier models get a short list of the hardest jobs. That's a problem for labs whose revenue forecasts assume unlimited growth. If you appreciate unpopular takes, follow me @m0xt_ for more.
172
296
103
2,768
375,360
Exciting news from ZenML. New version of Kitaru is the missing link to get value out of your agent traces.
This quoted post is unavailable.
1
1
4
195
Why is it called AWS IAM Identity Center when it's completely separate from IAM?
Finding CDK examples rather than AWS console instructions makes me happy
TIL: AWS ECS Fargate tasks triggered by EventBridge rules may fail despite the rule saying successful invocation. If so, the only way to detect this is to listen to the 'ECS Task State Change' event with stopCode TaskFailedToStart.
I wasn't wondering why the German policy was interested in Przybylski's star until I realized that those were separate tweets
1
A fun part of scaling a startup is recruiting. We need front end developers in Findable: findable.crew.work/jobs
1
Khaby Lame should see this! If I ever decide to do a tri, I’ll put my shoes on before getting on the bike.
Le départ du cyclisme est capital dans une épreuve de triathlon. La preuve en image. 🎥 fb Médéric Pion
1
I might dislike many tweets, however this one brings the taste of vomit. Is Mr Karp delusional or just selling?
Det har vel gått et par år, men AI er fortsatt mønstergjenkjenning uten at det trenger å være negativt
"I dag funker det meste av AI veldig dårlig, og det meste som kalles AI er kun avansert mønstergjenkjenning trent på store mengder data. Ekte AI vil lære av mindre data." <- @KnutHellan Lytt nå: hubs.li/Q010vz4W0 #ai
1
Strava down. Good thing I just had a chill ride on Zwift. If this had happened when I finished a challenging outdoors ride on Friday, I would have been stressed out #cyclinglife
4
This is how traffic lights should work in all urban areas.
"In a right-side up world, drivers would have to press a beg-button to cross the street..." The Dutch have it sorted. At this #Delft intersection, cyclists and pedestrians get a continuous green; the arrival of a car or truck sets off a sensor, which allows it to cross.
1
When I see a Python library using six, I get weary. Supporting Python 2 shouldn't be necessary anymore.
Knut O. Hellan retweeted
Open PhD position in Deep Reinforcement Learning at @OsloMet @OsloMetAI. Apply with your own project. I am available as supervisor for Neural CA for RL projects (category "Particularly innovative research ideas in DRL"). Let me know if interested ;) oslomet.no/en/work/job-openi…
16
20
Knut O. Hellan retweeted
Exciting PhD opportunity if you're interested in RL and NLP!
Interested in doing a PhD at @UCL_DARK? @egrefen and I are looking for strong&diverse applicants for UCL scholarships. Please email CV, personal statement, and research proposal to [email protected] by Dec 1. Interviews Dec 7-11. Lab site: dark.cs.ucl.ac.uk/
4
14
The M1 MacBooks look tempting. Sadly lacks a few mandatory tools so I'll wait a few months.
1