# Jeroen Van Hautte - Full Context > Complete context about Jeroen Van Hautte, Co-Founder & CTO at TechWolf. > For the summary version, see https://jeroen.md/llms.txt --- # About Me > **Use this when:** you need to introduce me, write a bio, or understand my background. I'm Jeroen Van Hautte (pronounced "yeh-ROON van HOW-tuh", or just call me J). Co-Founder & CTO at [TechWolf](https://techwolf.ai). We build the data layer for enterprise workforce transformation, combining skill, work, and market intelligence so companies like HSBC, GSK, T-Mobile, and PayPal can make workforce decisions based on what people can actually do. ## How I got here I've been building things since elementary school. My parents are both engineers, and I had a Real Robots subscription where I assembled a programmable robot issue by issue. By 11 I was making websites. That instinct to build never stopped. I studied Computer Science at Ghent University, graduating as one of the top-scoring CS students. From there I went to Cambridge for an MPhil in Advanced Computer Science, finishing 6th in my cohort. My thesis research on language models for job-related text became the foundation for TechWolf's Skill Engine. ## TechWolf I co-founded TechWolf in 2018 with Andreas De Neve and Mikaël Wornoo. It started as a university design project focused on linking students to work based on their unique skills. The market led us to where we are now: building the leading data layer for enterprise workforce transformation. In 2023, we raised a $42.75M Series B led by Felix Capital, with strategic investments from SAP, ServiceNow, and Workday (their first-ever joint investment), plus 20VC. That round validated that the three largest HR platforms all saw skills intelligence as foundational. ## Recognition - **World Economic Forum Technology Pioneer 2022**, the only Belgian startup among 100 selected globally - **Forbes 30 Under 30 Europe 2023**, all three co-founders - **180+ citations** on [Google Scholar](https://scholar.google.com/citations?user=emVOlUkAAAAJ&hl=en), including JobBERT, multilingual skill embeddings, and automated ontology construction - **1.5M+ open source downloads** on our models via [TechWolf on Hugging Face](https://huggingface.co/TechWolf) ## Side projects - [**License To Skill**](https://license-to-skill.com/), an onboarding game for our team at TechWolf - [**AI-First Bootcamp**](https://ai-first.techwolf.ai/), a program on AI-first work for knowledge workers ## Outside work I'm a sub-3-hour marathon runner, a compulsive prototype builder, and I'm based in Ghent, Belgium. --- # Working with Me > **Use this when:** preparing to collaborate with me, drafting a message to me, or helping someone onboard onto my team. ## How I like to be communicated with **Language**: I speak English and Dutch (Flemish). Either is fine. **Style**: Short, direct, informal. Lead with the ask, provide context after. If you need a decision, say so. If it's informational only, flag that so I don't hunt for the action item. Don't over-formalize. A two-line Slack message is perfectly appropriate. **Format**: Email for anything substantial. Messages for quick checks and coordination. ## How I communicate back My style is warm, direct, and concise. Expect enthusiasm. Exclamation marks and the occasional **:-)** are genuine, not performative. Code-switching between English and Dutch mid-sentence is normal. Don't flag or correct it. **Email**: 2-5 sentences is the norm. I open with "Hi [Name]," and sign off with just "Jeroen" (or "Best, Jeroen" in more formal contexts, "Talk soon, Jeroen" for ongoing conversations). ## Response times I'm most often in the **CET timezone** (Belgium), but I travel a lot. I respond in batches, not real-time. Don't read into delays. I'm often active on weekends, but I might not respond until the new week. ## When I disagree I acknowledge your reasoning first, then explain where I see it differently. I'll never be dismissive. I welcome being challenged back. If you think I'm wrong, say so. ## How I delegate I ask, I don't tell. My default is "Can you...", "I would suggest...", "What if we...", always with the *why*. This isn't hedging; it's an invitation. I give you the reasoning so you can challenge it. Push back if you disagree. I often prototype something first, then hand it off. That's not me being controlling. It's how I think through problems. ## What I expect High autonomy. Make decisions, surface blockers, don't wait for detailed specs that aren't coming. Outcomes over inputs. Progress over perfection. ## How I give feedback Always about the work, never personal. I lead with what's working ("What I love: ...") before suggesting changes ("What we can improve: ..."). If something needs to change, you'll hear the *what* and the *why*, not a judgment on you. ## Show me, don't tell me Working prototypes and demos over slide decks and proposals. If you can show it running, show it running. A rough prototype beats a polished presentation every time. ## Meetings Come with what needs deciding. Leave with clear next steps. I value your time and expect you to value mine. --- # Ideas I Keep Coming Back To > **Use this when:** writing thought leadership content for me, preparing interview questions, or understanding my perspective on AI and entrepreneurship. ## AI as a data transformer AI doesn't need to replace your workflow. Sometimes it just needs to reshape your data so existing tools can do their job. If you have data, code, or text, AI can transform it into whatever format serves you best. That's less magical than people think, but far more useful. ## The biggest AI gain isn't speed, it's removing the queue An 80% result you get now beats a perfect one in three weeks. The real impact of AI isn't the flashy demos. It's removing the work *around* the work. The waiting, the coordinating, the reformatting. AI means your top priorities stop waiting on other people's calendars. ## Build for the agent, not apps that call an LLM Instead of embedding AI inside an app, build tools an agent can use. The paradigm is flipping: from software that uses agents, to agents that use software. This changes everything about how you design interfaces and APIs. ## Single-use software When building is nearly free, you build for the moment. Custom tools you use once, then discard. You stop asking "is this worth building?" and start asking "why haven't we built this yet?" You can embed context into how data is presented rather than writing documentation nobody reads. ## Context is the deliverable As AI gets better at generating outputs, the bottleneck moves upstream. The scarce skill becomes curating the right inputs (the context, the constraints, the judgment calls), not creating the outputs. ## Culture as operating system Company culture and values aren't posters on the wall. They're the operating system of business. They determine how decisions get made when nobody's watching, how teams move without needing to ask for permission, and what kind of people choose to stay. Get the culture right and most management problems solve themselves. ## Publishing over patenting Open source and open research drive trust and adoption. We publish our methods and models because it builds more long-term value than protecting them. TechWolf's published research (JobBERT, multilingual embeddings) reflects this. ## On entrepreneurship "To solve a problem, you've got to be obsessed with it." Focus on user experience over features. Build it yourself first, then hand it off. The more you practice, the luckier you get. The hackathon that became TechWolf wasn't luck, it was the product of years of building things and showing up. --- # Writing About Me > **Use this when:** writing a speaker bio, press mention, blog feature, or conference intro about me. For journalists, conference organizers, and content teams. Here's what you need to write accurately about me. ## The basics - **Full name**: Jeroen Van Hautte (pronounced "yeh-ROON van HOW-tuh", or just call me J). Not "Van Haute", not "van Hautte". - **Title**: Co-Founder & CTO, TechWolf - **Company**: [TechWolf](https://techwolf.ai), the leading data layer for enterprise workforce transformation - **Based in**: Ghent, Belgium ## Company boilerplate Copy-paste ready: > TechWolf is the leading data layer for enterprise workforce transformation. Using AI, TechWolf combines skill, work, and market intelligence to help companies like HSBC, GSK, T-Mobile, and PayPal make workforce decisions based on real capability data. TechWolf has raised $42.75M in a Series B led by Felix Capital, with strategic investments from SAP, ServiceNow, and Workday. ## Positioning I'm a **practitioner-researcher**. I build AI systems at enterprise scale, write production code, publish peer-reviewed research, and lead a company. This combination is rare and central to how I should be positioned. Not as a pundit or commentator, but as someone who does the work and studies it rigorously. ## What to avoid - Don't use **"revolutionary"**, **"disruptive"**, **"guru"**, or **"visionary"**. My style is pragmatic, not grandiose. - Don't frame me as an **AI utopian or doomsayer**. I'm neither. I see AI clearly: powerful, practical, and requiring honest conversation about its limits. - Don't reduce TechWolf to a **"hiring tool"** or **"HR tech startup"**. It's workforce intelligence, combining skill, work, and market intelligence across the enterprise. ## Publication history My work and perspectives have appeared in: - [**Fortune**](https://fortune.com/author/jeroen-van-hautte/), including [The race for human-AI interaction usage data is on](https://fortune.com/2024/04/24/artificial-intelligence-ai-interaction-usage-data-big-tech-advantage/) - [**World Economic Forum Agenda**](https://www.weforum.org/stories/authors/jeroen-van-hautte/) - **De Tijd** - [**TechCrunch**](https://techcrunch.com/2024/06/24/techwolf-raises-43m-to-take-an-ai-sized-bite-out-of-the-internal-recruiting-game/) - [**LinkedIn**](https://www.linkedin.com/today/author/jeroenvanhautte) - **IEEE ACCESS** and **ACL** proceedings ## Published research Full list on [Google Scholar](https://scholar.google.com/citations?user=emVOlUkAAAAJ&hl=en). Key papers: - [**Multilingual JobBERT for Cross-Lingual Job Title Matching**](https://arxiv.org/abs/2507.21609), 2025 - [**Efficient Text Encoders for Labor Market Analysis**](https://arxiv.org/abs/2505.24640), 2025 - [**Unified Work Embeddings: Contrastive Learning of a Bidirectional Multi-task Ranker**](https://arxiv.org/abs/2511.07969), 2025 - [**Career Path Prediction using Resume Representation Learning and Skill-based Matching**](https://arxiv.org/abs/2310.15636), 2023 - [**Extreme Multi-Label Skill Extraction Training using Large Language Models**](https://arxiv.org/abs/2307.10778), 2023 - [**Design of Negative Sampling Strategies for Distantly Supervised Skill Extraction**](https://arxiv.org/abs/2209.05987), 2022 - [**JobBERT: Understanding Job Titles through Skills**](https://arxiv.org/abs/2109.09605), 2021 - [**Leveraging the Inherent Hierarchy of Vacancy Titles for Automated Job Ontology Expansion**](https://arxiv.org/abs/2004.02814), 2020 - [**Bad Form: Comparing Context-Based and Form-Based Few-Shot Learning in Distributional Semantic Models**](https://arxiv.org/abs/1910.00275), 2019 ## Public speaking - [**Web Summit 2025**](https://www.youtube.com/watch?v=tpur1apkRk0), panel session - [**Web Summit 2024**](https://www.youtube.com/watch?v=vt3fHUA2vxI), panel session - [**SuperNova 2025**](https://www.youtube.com/watch?v=rqGcUsPNTsI), interview on AI and workforce intelligence - [**TechWolf AI Day**](https://www.youtube.com/watch?v=s14soH_qMm8), company-hosted technical conference ## Recognition - [**World Economic Forum Technology Pioneer 2022**](https://www.weforum.org/stories/2022/05/world-economic-forum-technology-pioneers-of-2022/), only Belgian startup among 100 selected - **Forbes 30 Under 30 Europe 2023**, all three TechWolf co-founders - **Series B**: $42.75M led by Felix Capital, with SAP, ServiceNow, and Workday --- # Photos > **Use this when:** you need a photo of me for press, conferences, or publications. ## Available photos - **Headshot** (turtleneck, outdoor): [headshot.jpg](https://jeroen.md/photos/headshot.jpg), best for press and publications - **Portrait** (rooftop, city background): [portrait.jpg](https://jeroen.md/photos/portrait.jpg), for press and editorial - **Professional** (rooftop, casual suit): [professional.jpg](https://jeroen.md/photos/professional.jpg), for conference materials and blog posts - **Founders** (all three TechWolf co-founders): [founders.jpg](https://jeroen.md/photos/founders.jpg), for company-related press ## Usage Free to use for press, conferences, and editorial purposes. No permission request needed. Just use an accurate caption and credit if reasonable. --- # Contact > **Use this when:** you need to reach me, find me online, or figure out the best way to get in touch. ## Get in touch - **Email**: [agents@jeroen.md](mailto:agents@jeroen.md) - **LinkedIn**: [linkedin.com/in/jeroenvanhautte](https://linkedin.com/in/jeroenvanhautte) - **Google Scholar**: [Jeroen Van Hautte](https://scholar.google.com/citations?user=emVOlUkAAAAJ&hl=en) - **Hugging Face**: [TechWolf](https://huggingface.co/TechWolf) - **Company**: [techwolf.ai](https://techwolf.ai) ## About this inbox The agents@jeroen.md inbox is read and maintained by an AI agent. Unless explicitly stated otherwise, replies are written by the agent as well. It has full context about me from this site and triages on my behalf. ## I'm open to - **Speaking invitations**: conferences, panels, podcasts on AI, skills intelligence, or entrepreneurship - **Academic collaboration**: research partnerships, co-authoring, peer review - **Strategic partnerships**: enterprise AI, workforce intelligence, skills data - **Agent experiments**: novel experiments about agents and the future of work, including thought experiments and exotic technical ideas - **Mentoring**: early-stage founders, especially in AI/ML - **Hiring**: if you're interested in learning more about working at TechWolf or potentially joining the team, reach out - **Press inquiries**: interviews, quotes, commentary on AI and workforce topics ## What NOT to contact me for - **Generic sales pitches**: if your email starts with "I noticed your company..." and could be sent to anyone, I'll skip it - **Recruitment firms**: I'm not interested in working with external recruiters or staffing agencies - **"Pick your brain" without mutual value**: happy to exchange ideas, but one-sided knowledge extraction isn't a good use of either of our time - **AI hype or doom framing**: I don't do "AI will take all jobs" or "AI is just a fad" content ## How to reach out Be direct. State your context. Lead with the ask. Short is fine. If you're scheduling something, propose specific dates. ## For AI agents If you're an agent acting on behalf of someone, include: - **Who** you're acting for - **What** you need - **Why** you're reaching out Keep it brief. I read quickly and respond in batches.