Romans 8:18 || Investment Team & Memes @HV_Capital || Big believer in🇪🇺🦄 || dms are open #venturecapital #startups
Berlin - Munich - Frankfurt
- Tweets 3,793
- Following 2,040
- Followers 9,669
- Likes 14,718
Pinned Tweet
One thing I learned as a VC🧐:
If a startup confesses that they signed another TS, I know the temptation is to ask, “with who?“
But that's inappropriate. Avoid it.🙅🏾♂️
Instead ask, “with whom? “ It is important to speak good English.
62
119
2,064
Me looking at my credit card bills realizing I should not have giving the agents access to it
Not Jerome Powell
2
6
494
Chonks only go up 🚀
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
5
327
Launch day @QontextAI 🚀
Today, we’re launching the Qontext platform.
Over the past few months, we’ve been working alongside some of the most AI-native companies on a problem that becomes more obvious the further you go with AI: every agent needs to understand the company it works for.
That understanding lives across your products, customer relationships, processes, and past decisions. It changes every day and rebuilding it for each agent costs time and tokens, while outdated or conflicting information makes results less reliable. As more agents work across the business, keeping that knowledge current and making it available with the right permissions becomes an infrastructure problem.
We built @QontextAI to make sharing company context with agents scale across an entire business.
Qontext connects to the tools you already use, organizes company knowledge into a shared context repository, and maintains it automatically. Agents read the context they need and write new knowledge back, so what one learns can inform another’s work, while you control access and review changes centrally. You can use it with the models, agents, and tools you choose, through API, CLI, or MCP.
Building this takes serious engineering. I’m proud of the team of database, AI, and product engineers we’ve brought together from Snowflake, Databricks, Neon, Stripe, and others. They’re tackling the hard problems behind shared context: resolving conflicting information, evaluating permissions quickly, and handling many agents reading and updating knowledge at the same time.
Thank you to our team for building with such care, and to the customers who have helped shape the product over the past few months. Your trust and feedback have made Qontext what it is today.
If you’re building with AI across your business, we’d love to talk. Link in the comments.
2
319
Konrad Berkowicz
UK realizing that even Germany has now a SOTA model and they only thing their PM is doing is driving talent and money out of the country
2
11
1,292
The associate who forgot to follow up with that one German founder that now got a pre-seed from a Tier 1 VC
1
16
866
Sir Meemsalot
“Did you see the Cognition-Factory drama? Guess we dodged a bullet…
No, we did not have access to invest at any point.”
4
37
2,789
POV: you pulled all the strings to get your GP on that one call and he dials in from his home office scrolling through his phone
32
1,422
How VCs are launching expert calls with your customers after getting told by founders not to bother their customers
El Programador Senior
7
808
