Maximilian Ahrens
@_MaxAhrens_
Followers
227
Following
217
Media
5
Statuses
66
Cofounder & CEO @maihem_ai (YC W24) | PhD in AI, Oxford | he/him
San Francisco | London
Joined August 2015
š¦¾š¤See you today at @Moonfire_VC's Pulse Summit to chat about adoptable AI robotics and governance for complex AI systems. Side note: London is absolutely buzzing this week with AI, robotics, and tech events! š«¶š¬š§ Follow us for more @maihem_ai
2
3
5
Today at @FabtechExpo : 4pm āļø š° Join me for Coffee, Cake, and Weld at the Starbucks by booth B12000. Iāll cover the tab. Bring your questions, your ideas, or just your curiosity about autonomous welding. You canāt miss me, Iām the guy in THAT t-shirt.
0
1
2
ā¹ļø The resonance is real. Our material has been flying off the table. What struck me most: manufacturers are visibly relieved to see AI software that makes welding robots autonomous without the need for programming. That pain point is universalāand it shows.
0
1
1
Heading into Day 2 at FABTECH Expo ā first impressions from yesterday: š¤š¦¾ AI-powered robotics is everywhere. Particular focus lies on technology that makes robot usage more versatile. Our AI software for adaptive welding robots does just that. @maihem_ai @FabtechExpo
1
0
1
In LLM land, a slow model is annoying. In robotics, a slow model can be disastrous! Visible pauses at best, dangerously jerky motions at worst. But large VLAs are slow by nature. What can we do about this? An in-depth š§µ:
12
62
457
"Robotics is going to be the biggest industry in the history of the planet" š¦¾ Robotics and automation will be the foundation of every factory that wants to scale. Hundreds of machines, guided by human oversight, working behind the scenes, delivering products faster, better,
7
12
91
Excited about the progress in open-source AI robotics! We can't wait to try it out in our AI robots.
Our vision is for AI that uses world models to adapt in new and dynamic environments and efficiently learn new skills. Weāre sharing V-JEPA 2, a new world model with state-of-the-art performance in visual understanding and prediction. V-JEPA 2 is a 1.2 billion-parameter model,
0
1
2
Last year's Pulse summit was already a fantastic gathering point for folks curious and courageous to shape the future. Mattias Ljungman, Mike Arpaia, Akshat Goenka and Alice Pillar from Moonfire made this year's edition even more exhilarating. Wouldn't want to miss it.
0
0
0
Excited to be joining an exceptional line-up at Moonfire Pulse 2025Ā to talk about how we designĀ adaptable AI systemsĀ for the physical world.
1
0
3
If youāre exploring how AI-powered robotics could unlock substantial productivity in your industry, DM me. Come join the conversation!
0
0
0
AI-powered robots are critical for reindustrializing economies, iterating faster on physical products, and staying globally competitive.
1
0
0
Dear DPD UK and @Geopost_news , we'd be more than happy to help you out! Last week was a tough one. How about a 50% discount on our solutions if you let us help robustify your conversational AI? We mean it š¤ ā¤ļø ! DM us! https://t.co/wP2X8WXLgf
bbc.com
The parcel delivery firm says the mistake was a result of a system update, which has been disabled.
0
1
1
Choose https://t.co/gtZn3BFwqD, not AI mayhem.
maihem.com
Adaptable AI robots for your complex and dynamic workflows
0
0
0
At @maihem_ai , we provide automated and comprehensive stress-tests for LLM-based applications with our state-of-the-art synthetic data simulations ā to make AI safer and perform better. Let us find the weaknesses of your AI before your customers do.
1
0
0
A case in point of what can happen if you skip sufficient quality assurance and robustness testing for your conversational AI.
1
0
0
āUtterly uselessā and āswears at customerā. No company wants such descriptions associated with its customer support AI ā let alone be prominently featured in the press for such an AI blow-up. Yet, thatās exactly what happened to @Geopost_news 's DPD...
bbc.com
The parcel delivery firm says the mistake was a result of a system update, which has been disabled.
2
0
1
This is joint work with my fantastic co-authors @d_erdemlioglu (IESEG), @mcmahonecon (Oxford University),Chris Neely (Federal Reserve Bank of St. Louis), and Xiye Yang (Rutgers University).
0
0
1
ā”ļøFinally, we find that speech-implied news - both good and bad news - increase volatility and tail risk. These results challenge the conventional view that central bank communication primarily resolves uncertainty.
1
0
1
š”We also find that financial markets attend to these news more closely during 'abnormal' GDP and inflation regimes.
1
0
1
š”We find that monetary policy news conveyed in US central bankersā speeches can help explain volatility and tail risk in both equity and bond markets. This is measured on high-frequency intraday US market data.
0
0
2