Caught up with our good friend Jeroen Arts from Speedinvest yesterday in Paris 🇫🇷, and he asked a fascinating question about how AI is reshaping the tutoring space.
Here are a few takeaways from that conversation (not gospel truths, just personal reflections):
First question I always ask when evaluating an AI tutor startup:
“If GPT‑5 launches tomorrow, does this product still need to exist?”
We all know LLMs are only getting smarter and broader. So for AI tutors to stay relevant, I look for signals like:
✅ Deep verticalisation:
Subject-specific mastery in areas like maths, reading, languages. LLMs will dominate general tutoring so niche depth wins over generic chatbots.
✅ Laser focus on core use cases:
Great tutors don’t try to do it all. I like products that solve one painful problem extremely well (think: oral fluency, exam prep, maths explanations). Bonus points for habit-forming UX -> à la Duolingo!
✅ Distribution where the learning happens:
Integrated into WhatsApp, LMS, school OS… I rarely get excited by standalone apps that add friction to a learner’s workflow.
✅ Agentic UX:
Tutors that DO things: guide, quiz, plan, remind, adapt, celebrate milestones. Not just passive Q&A machines.
✅ Learning loops built in:
Systems that get smarter with feedback from real student interactions (voice, handwriting, usage patterns, etc.).
—
🚩 Easy pass for me when I see:
– GPT wrappers with no UX value or subject-specific expertise
– No credible moat if Big Tech releases a better tutor tomorrow
– No clear link to outcomes, no pedagogical frameworks, no alignment with local K12 curriculums
—
Just sharing my lens and of course, I might be wrong as this is an evolving process.
How AI is reshaping the tutoring space (LinkedIn post)
In