We've spent a decade hacking growth. The next decade will be about engineering it. At Web Summit I talked about what that shift really means.

At @MAI_agents, we didn't add AI to existing ad tools. We built autonomous agents that own the entire workflow. That's the difference between a feature and a company. We don’t treat ‘AI-native’ as a buzzword, it's our strategy. See for yourself —>

We're in the brief window where 'AI-native' is still an advantage. But soon it'll be table stakes. Just like mobile-first. The question: are you rebuilding from scratch, or patching the old system?

Companies that bolt on AI still have: Legacy systems, manual workflows, and human-centric processes. While AI-native companies move faster, scale cheaper, and learn continuously. This creates a moat.

AI-native doesn’t mean better features. It’s a different architecture. Your data model, your UI, your entire product logic, It all assumes an agent is doing the work, not assisting a human.

Bolt-on AI is the same trap. You're adding AI features to workflows designed for humans. AI-native means rethinking the entire process around what agents do best: Tireless execution, pattern recognition, and real-time optimization

Remember when companies tried to bolt mobile onto desktop websites? Clunky, slow, bad UX. Mobile-first companies like Instagram and Uber designed everything around the phone. No surprise they won.

‘AI-native' is becoming the new 'mobile-first.’ Companies building with agents from day one will have insurmountable advantages over ‘bolt-on’ AI. Let me explain…

The marketplace is too dynamic for one-size-fits-all rules anymore. You need systems that make precise decisions every hour based on real-time data. That's what engineering growth actually means. See how we do it at https://t.co/sKDxNc7XkF.

AI agents now change this: • They collect data • Reason about improvements • And implement changes without you writing code. The growth flywheel is now accessible to everyone.

Building these systems used to require armies of data engineers, ML engineers, and product engineers. The cost of hiring and maintaining that team was completely out of reach for most businesses. So everyone else was stuck guessing.

Big Tech figured this out years ago. Netflix doesn't guess what you'll watch It builds feedback loops that learn from every click and improves automatically. Same with search engines and social feeds.

You can’t optimize what your system doesn’t see. Ads, Analytics, and @Shopify don’t talk, so the system chases purchases instead of profit. It might look like growth, but you’re optimizing for the wrong things. With the wrong goals, or incomplete data, your system might be

8/ If you enjoyed this, follow me @yuchen_wu for more. I’m sharing weekly threads on what we’re learning while building an agentic marketing platform from the ground up. You can find us at https://t.co/sKDxNc7XkF

7/ Competition is at an all time high. Web Summit reminds you there’s hundreds of teams solving the exact same problem you are. So it’s really important to not only solve it really well, But to do it while running as fast you can :)

6/ No single company dominated the conversation. Foot traffic was spread across hundreds of booths. Every stage was filled with people. With no clear winner, this told me we’re still early with AI.

5/ AI agents have gone global. 2700+ startups from US, Europe, Asia - every section packed. Global momentum with AI was palpable.

4/ Marketing and media had a lot of buzz within Creative summit. What stood out: not everything was about AI. A lot of talk about how media is shifting away from legacy ways of thinking: • Where the creative comes from • What powers campaigns now • And what role does AI

3/ There were two types of AI demos on the floor: • Novel agentic apps unlocking things that weren't possible before • Existing software with AI layered on top The difference between the two was obvious walking booth to booth. Agentic apps captured more attention while it

2/ We're past the "is this real?" phase with agents. Startups targeting niche verticals are the ones to watch. There are so many industries underserved by tech or stuck with legacy SaaS. Any one of these feel like the faster path to real revenue.