Philippe Mizrahi is the CEO and Co-Founder of Linkup, a Paris-based startup building the web search layer for the AI era. Previously a Group Product Manager at Lyft, Mizrahi co-founded Linkup in 2024 alongside Denis Charrier, whose prior company Niland — one of Europe's first vector search engines — was acquired by Spotify, and Boris Toledano (ex-McKinsey). The company has raised over $10M in funding, including a seed round led by Gradient, and is backed by Seedcamp, Motier Ventures, and angel investors including founders from Mistral, Datadog, and Deel. Linkup's API powers AI agents at enterprise clients including KPMG, and holds state-of-the-art results on OpenAI's SimpleQA benchmark.

AGENDA:

• 00:00:55 - Phil Mizrahi and the bet that became Linkup

• 00:04:07 - Why Linkup's original vision was wrong

• 00:07:00 - The MVP mistake most founders never catch

• 00:10:05 - Fundraise or bootstrap: what Linkup chose and why

• 00:13:08 - What Phil looks for in a founding team

• 00:15:50 - Why Linkup wins in a crowded AI market

• 00:19:07 - How Linkup got its first customers

• 00:21:55 - The market bet Linkup is building toward

• 00:36:22 - The pricing psychology behind Linkup's strategy

• 00:39:05 - Why most startups target the wrong customer

• 00:42:07 - The growth loop that scaled Linkup

• 00:44:40 - Running a global team before you're ready

• 00:46:45 - What separates founders who execute from those who don't

• 00:50:09 - What most founders still get wrong about AI

• 00:53:25 - How AI rewrites the zero-to-one playbook

• 00:56:36 - What Phil tells every early-stage founder

Most founders using AI today are building the same thing.

Same prompts. Same outputs. Same products.

So the real question isn't "How do I use AI?"

It's "How do I stay differentiated when everyone has access to the same models?"

This week on Zero to One, we sat down with Olivier Toubia, professor at Columbia Business School and a leading researcher on AI, innovation, and entrepreneurship.

💡 Some of the themes we explored:

🎙️ AI increases productivity, not differentiation

🎙️ Your unique context is the moat

🎙️ The bottleneck has shifted to execution and distribution

🎙️ Human judgment matters more than ever

🎙️ AI is a tool, not the strategy

This episode is worth your time if you’re building in AI and trying to figure out what actually creates a MOAT in the next generation of startups. 🚀

Most AI startups die before they even ship.

Worse: they spend months building the wrong thing.

This week on Zero to One, Ankur Patel (Multimodal / AgentFlow) dropped some of the clearest, most brutal (and useful) advice we’ve heard for anyone trying to go from zero to one in enterprise AI. 💡

5 hard truths for builders:

1. Speed is your only moat Foundation models evolve weekly. If you’re not shipping fast, you’re irrelevant. Velocity beats architecture, funding, and cleverness.
2. Pick your wedge and obsess Startups that try to be a platform from day one don’t make it. Start wide → feel for the pull → double down on 1–3 use cases people will pay for.
3. Trust is more important than tech Enterprises won’t deploy agents that feel like black boxes. You need: traceability, sandbox validation, human-in-the-loop checks, clear ROI. No trust = no production.
4. Being an outsider is an edge If you’re young and don’t know the rules of the industry, good. You’ll move faster and see opportunities incumbents overlook.
5. Failure is leverage now Failing at a startup doesn’t hurt your career anymore, it accelerates it. You’ll either win or become 10x more valuable.

This episode is worth your time if you’re building in AI and trying to get real traction.

What’s your biggest blocker going from zero to one? Drop it in the comments.

#startups #AI #founders #zerotoone #enterpriseAI #agenticAI #podcast