5/ Ultimately, the move toward multi-agent platforms signals a future where AI systems collaborate seamlessly, delivering more tailored, impactful experiences for every user wherever they work.

4/ Agents Everywhere:  Their vision is to create a platform of AI agents that work seamlessly within users’ existing tools—email, documents, chat, and more. Sometimes you want agents embedded seamlessly in your workflow, following you from app to app. Other times, you need a.

3/ From Single Agent to Multi-Agent: Grammarly’s journey began with a single “agent”—the grammar assistant—delivering real value at scale before evolving into a broader platform where third parties can build and deploy their own agents, similar to an app marketplace.

2/ Building an Ecosystem: The ultimate aim is to create a stable of best-in-class AI brands that work together, much like how Google owns YouTube or Meta owns Instagram and WhatsApp. Coda (another company previously acquired by Grammarly) and Superhuman are key parts of this.

1/ I love seeing companies like Grammarly competing with the largest players in tech. The recent convo on @twistartups about their acquisition of Superhuman is a must-listen. Here are a few ideas that stood out 👇.

If the AI Act is making it difficult for companies with huge legal departments to do business in the EU, then what do you think it will do for startups?.

The Digital Services Act (DSA) gives users more transparency, but also gives states new powers to mandate content removal in the EU. What is everyone's take on it?.

5/ This trip really highlighted for me how digital control can leapfrog traditional development. Even in places where basic infrastructure lags, governments can implement sophisticated surveillance systems quickly. It’s a vivid example of how the future isn’t evenly.

4/ Kyrgyzstan is technically the only democracy in its region, but the “big brother” atmosphere is palpable. Since the pandemic, the government has increased surveillance and limited free speech, following a pattern seen in many neighboring countries (read about Turkmenistan!).

3/ Cameras are everywhere, and the police presence is significant. During my two-week stay, I received three traffic tickets, all of which showed up almost instantly in a real-time online portal. Paying fines is streamlined with kiosks in hotels and airports—you apprarently can’t.

2/ On one hand, the country’s physical infrastructure is still developing—unfinished roads, rural areas with yurts and no electricity, and a strong Soviet legacy in the architecture and culture. Yet, in the middle of all this, the digital infrastructure for surveillance is.

1/ I went to Kyrgyzstan recently on holiday. The drives and hikes were really incredible (story for the another day), but what also struck me was the country’s ongoing transformation into a surveillance state.

🙏🍿 War for the Communication Layer. Would love to see some real disruption in the space. Too much of whats happening in AI is the biggest companies leading the way.

5/ Ironically, as AI and LLMs take over, we’re rediscovering the value of structured, comprehensive docs. To train an AI, you need the modern equivalent of a manual. Is it time to bring manuals back—not just for humans, but for the machines that help us learn?.

4/ But something’s lost: Manuals told the whole story, not just isolated facts. Without them, users often miss the big picture or the “why” behind the product.

3/ With software updating constantly, static manuals are obsolete before they’re printed. Docs are now living documents—wikis, release notes, and community forums.

2/ Why did they disappear in the first place? Users want answers now, not hours of reading. Searchable FAQs, tooltips, and instant support have made manuals feel slow and overwhelming.

1/ Remember when every product came with a thick, definitive manual? Those days are gone—manuals quietly faded away, replaced by bite-sized, on-demand help. But what if they’re on the verge of making a comeback? 👉🧵.

Legacy education: ‘Show me your work.’ .Modern education: ‘Show me your prompt.’.

On our most recent Trillion Agents podcast, we dive deep into what’s really moving the needle in the AI agent space. From the evolution of context engineering to why vertical AI startups are pulling ahead, here’s what we unpacked:. 1. Why context engineering is more important.