🌊 AI Agent Directories

AI Factories, AI-native go-to-market tech stack, Microsoft’s Ambient AI, Anthropic’s guide to AI Agents & OpenAI's Robotics

👋 Hello! I’m Ivan, vc and founder. I share curated research from the top 1% of founders and investors—helping you focus on what matters most in tech.

Surf emerging tech waves with 15K+ startup riders:

---

Summary

1. 🌊 AI Factories

2. 🌊 AI Agent Directories

3. 🌊 AI-native go-to-market tech stack

4. 🌊 Microsoft’s prophecy

5. 🌊 Anthropic’s guide to AI Agents

6. 🌊 NVIDIA’s product roadmap

7. 🌊 OpenAI moves into Robotics

8. 🌊 AI’s shift from training to inference

9. 🌊 Founder Compensation 2024

10. 💰 Recent Deals

11. 🍫 Meme-ology

Want to sponsor Startup Riders? Fill out this form.

---

This newsletter is proudly sponsored by Yellow Birds Creative.

Creating organic content for your socials is tough. It's even tougher to do so in a way that accurately reflects your brand identity. We offer 3 packs:

• Content creation: photography, video, video podcasting, copywriting.

• Social Media Management: taking care of all your content needs.

• Project Management: ideate, plan, and manage marketing initiatives.

---

🌊 AI Factories: The Race Is On

Hyperscalers’ CapEx is set to hit $𝟮𝟰𝟰𝗕 𝗶𝗻 𝟮𝟬𝟮𝟱 - this is huge.

What does this mean? It means scaling isn’t dead—it’s transforming.

Big tech is building AI-first infrastructure: multi-gigawatt data centers, high-bandwidth networks, and cutting-edge hardware designed for training and inference at scale. Here’s why:

• AI workloads demand massive compute power and efficiency.

• Companies like NVIDIA, with GPU racks and memory-optimized systems, are enabling hyperscalers to push AI capabilities further.

• Synthetic data and inference reasoning are fuelling innovation, creating new demands for AI compute.

---

🌊 AI Agent Directories

I recently wrote about AI Agents (start there if you don’t have context).

Here are 3 great resources to track the market:

1. Built by Dharmesh Shah: agent.ai/agents

2. Built by Nikolas Barwicki: aiagentslist.com

3. By AI Agents Directory: aiagentsdirectory.com

Source

---

🌊 AI-native go-to-market tech stack

Source

Found a great visualisation by Battery Ventures on something I’ve been thinking a lot about lately - how can startups leverage AI GTM tooling?

In case you missed it - on the past, present and future of sales:

1. 🌊 𝗦𝗮𝗹𝗲𝘀 𝗧𝗲𝗰𝗵 𝗦𝘁𝗮𝗰𝗸

2. 🌊 𝗦𝗮𝗹𝗲𝘀 𝗔𝗜

3. 🌊 𝗔𝗜 𝗳𝗼𝗿 𝘀𝗮𝗹𝗲𝘀 𝗶𝗻 𝟮𝟬𝟮𝟱

---

🌊 Microsoft’s Prophecy: Ambient AI

Source: Podcast episode from the great BG2 podcast by Bill Gurley and Brad Gerstner on Youtube.

Microsoft's CEO says AI agents will transform SaaS.

Here’s what that means:

• 𝗧𝗿𝗮𝗱𝗶𝘁𝗶𝗼𝗻𝗮𝗹 𝗦𝗮𝗮𝗦 𝗮𝗽𝗽𝘀 𝗮𝗿𝗲 𝗷𝘂𝘀𝘁 𝗖𝗥𝗨𝗗 (𝗰𝗿𝗲𝗮𝘁𝗲, 𝗿𝗲𝗮𝗱, 𝘂𝗽𝗱𝗮𝘁𝗲, 𝗱𝗲𝗹𝗲𝘁𝗲) 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲𝘀 𝘄𝗶𝘁𝗵 𝗿𝘂𝗹𝗲𝘀: essentially fancy (and not so fancy) interfaces (like Salesforce, Asana, or Notion) sitting on top of databases where users input and manage data with some extra features (business logic).

• 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁𝘀 𝘄𝗶𝗹𝗹 𝘁𝗮𝗸𝗲 𝗼𝘃𝗲𝗿 𝘁𝗵𝗲 “𝗿𝘂𝗹𝗲𝘀” 𝗽𝗮𝗿𝘁 (𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗹𝗼𝗴𝗶𝗰): Instead of rules being hardcoded into each app (e.g., Salesforce automating workflows or permission settings), AI will dynamically manage those rules across multiple apps or databases. For example: An AI agent could pull data from Salesforce, update a Notion page, and send a Slack notification—all at once.

• 𝗔𝗜 𝘄𝗶𝗹𝗹 𝘀𝘁𝗼𝗽 𝗰𝗮𝗿𝗶𝗻𝗴 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗯𝗮𝗰𝗸𝗲𝗻𝗱 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲: Right now, each SaaS app works with its own database. In the future, AI agents will work across many databases without worrying about the specifics of their backends (e.g., it won’t matter if one uses SQL and another uses MongoDB).

• 𝗕𝗮𝗰𝗸𝗲𝗻𝗱𝘀 𝘄𝗶𝗹𝗹 𝗯𝗲𝗰𝗼𝗺𝗲 𝗿𝗲𝗽𝗹𝗮𝗰𝗲𝗮𝗯𝗹𝗲: If all the “smart” stuff happens at the AI layer, the underlying SaaS apps and databases become less important. Companies might switch backends (or replace apps entirely) because the AI can adapt seamlessly.

• 𝗧𝗵𝗲 𝘀𝗵𝗶𝗳𝘁 𝘁𝗼𝘄𝗮𝗿𝗱 𝗔𝗜-𝗻𝗮𝘁𝗶𝘃𝗲 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗮𝗽𝗽𝘀: Businesses will demand apps built from the ground up to work with AI agents, rather than retrofitting AI onto old CRUD-based systems.

• My 2 cents: While this vision is compelling, the timeline may be longer than they think. True "AI-native" systems need both seamless integration across fragmented software ecosystems and significant advancements in AI's ability to understand nuanced business logic. Legacy systems won't disappear overnight, and adoption in enterprises—where SaaS is deeply entrenched—could take years.

• The opportunity? Founders who build modular, AI-first apps today are positioning themselves to lead when the shift happens.

---

🌊 Anthropic ($40B) dropped insights on building AI agents.

Source

A quick actionable summary from their post:

• 𝗔𝗴𝗲𝗻𝘁𝘀 𝗩𝘀 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀: Workflows are like GPS apps with fixed routes. Agents are like drivers who adapt to roadblocks and make their own decisions.

• 𝗦𝘁𝗮𝗿𝘁 𝘀𝗶𝗺𝗽𝗹𝗲: Most tasks don’t need fully independent agents. Start with workflows (like a checklist) and add complexity only when needed.

• 𝗦𝗺𝗮𝗿𝘁𝗲𝗿 𝘀𝘆𝘀𝘁𝗲𝗺𝘀: Memory lets AI “remember” past interactions, and vector databases help it retrieve the right info quickly—like a super-focused Google.

• 𝗧𝗼𝗼𝗹𝘀 𝗺𝗮𝘁𝘁𝗲𝗿: Agents rely on tools to act. Clear instructions and smart design make the difference between success and chaos.

• 𝗪𝗵𝗲𝗻 𝘁𝗼 𝘂𝘀𝗲 𝗮𝗴𝗲𝗻𝘁𝘀: They’re ideal for messy, open-ended tasks—like coding, research, or complex troubleshooting—where you can’t map out every step.

• My 2 cents: Anthropic’s advice is clear: Keep it simple. Focus on outcomes, not flashiness. The future belongs to reliable, impactful AI agents that just work.

---

🌊 NVIDIA ($3.6T) released their roadmap

Source

Jensen Huang - wearing a brand new leather jacket - just shared his roadmap on what comes after AI agents. Its worth a watch, but go straight to 1:15:45.

As a counter-point to this, I recommend you read this article by Harimus on the “Common misconceptions about the complexity in robotics vs AI”, tl-dr:

• Robotics and AI face different challenges. While AI thrives in static, data-rich environments, robotics battles the unpredictability of the real world, where sensorimotor tasks like grasping or balancing (Moravec’s Paradox) are much harder than abstract reasoning.

• Current AI models aren’t fast or precise enough for real-time robotics, where mistakes can’t be laughed off like a ChatGPT error.

• Despite promising efforts like Large Behavior Models, progress in robotics requires a different playbook—one that factors in the chaotic complexity of the physical world and moves beyond the hype.

---

🌊 𝗔𝗜’s 𝘀𝗵𝗶𝗳𝘁𝗶𝗻𝗴 𝗳𝗿𝗼𝗺 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 → 𝗶𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲

Best podcast I’ve heard all year on AI by Invest Like The Best with Benchmark’s Chetan Puttagunta. Here’s the tl-dr:

• 𝗔𝗜 𝗶𝘀 𝘀𝗵𝗶𝗳𝘁𝗶𝗻𝗴 𝗳𝗿𝗼𝗺 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝘁𝗼 𝗶𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲: The focus is now on using models effectively, with costs aligned to usage rather than massive upfront investments. Think of it like cars—few companies build engines from scratch; the real innovation happens in design, performance, and user experience.

• 𝗕𝗮𝗿𝗿𝗶𝗲𝗿𝘀 𝘁𝗼 𝗲𝗻𝘁𝗿𝘆 𝗮𝗿𝗲 𝗳𝗮𝗹𝗹𝗶𝗻𝗴: Rapidly evolving foundation models like GPT or Llama mean challengers—both horizontal (broad tools) and vertical (niche solutions)—can compete without needing to build their own models.

• 𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗶𝘀 𝘁𝗵𝗲 𝗻𝗲𝘄 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁𝗶𝗮𝘁𝗼𝗿: Success now hinges on creating 10x better applications and owning direct customer relationships, rather than just owning the largest or most advanced models.

Source: 📌 Source: Patrick OShaughnessy's "Invest Like the Best" Podcst, Episode = "Capital, Compute and AI scaling" ft Chetan Puttagunta.

---

🌊 OpenAI moving into Robotics

‘General-purpose robots that operate in dynamic real-world settings,’ with plans for a ‘wide variety of robotic form factors.’

— OpenAI Job listing

Looks like Caitlin (ex-meta AR), is hiring for robotics hardware roles at OpenAI:

---

🌊 Founder Compensation

Creandum just released their 2024 Founder Compensation survey results.

Check out their founder compensation calculator.

Source

---

🐇 Follow the white rabbit 🕳️

Other insights I’ve recently published:

• OpenAI’s o3 model announcement in a nutshell

• Incumbents vs. Startups in AI Agents

• Horizontal Vs Vertical AI: Where is the market going towards in 2025?

• The 2025 IPO market is heating up

• Enterprise LLM Adoption in 2024

• Data is the fossil fuel of AI by Ilya Sutskever

• The Modern AI Stack by Manoj Gupta

Other interesting reads:

• The Inference Landscape by Eric Flaningam

• LLMs for Dummies by Rex Woodbury

• How Large Language Models work by Andreas Stöffelbauer

• Large language models, explained with a minimum of math and jargon by Timothy B. Lee and Sean Trott

• Daniel Dines, UiPath CEO & Founder | E1240 by 20VC

• The Post Individual by Yancey Strickler

• Common misconceptions about the complexity in robotics vs AI by Harimus Blog

---

💵 Recent Deals

Southern Europe’s startup ecosystem is thriving.

Here’s where we’ve invested in 2024: Across defence, cybersecurity, robotics, AI agents, health-tech, wealth-tech, fintech, and logistics.

• Equixly API Security | API security testing platform

• GPTadvisor | Wealth Management gen-AI platform

• Genesy | Data aggregation and sales automation

• Chainloop | Automating trust for Software Supply Chain

• Cofers | Cash flow management / bank sync

• Galtea | Compliant and reliable GenAI development

• Angstrom | GenAI-based molecular simulations

• THEKER Robotics | AI Robotics for industrial processes

• Omniloy | Uncovering customer insights with AI

• Hoola | AI Agents for e-commerce

• Find Balance | Personalised GLP-1 companion

• Rauda AI | AI platform for customer service

• PuntoPost | Out-of-home delivery services

• XRF | Decision making solutions for complex scenarios

---

🍫 Meme-ology

---

Thanks for reading!

If you enjoy Startup Riders, I’d really appreciate a share - see you next month! 🤙

---

About Me:

I’m a VC, founder, surfer and BJJ purple belt—writing about tech, startups, and venture capital. I've built products at Facebook, helped grow a unicorn at Bloomreach, and bootstrapped Revenue Squared to six figures. I back founders (€200K-2M) and share curated research on Startup Riders.