On July 21, Pavel Durov announced that Gram Wallet would be integrated into every Telegram client this summer. A non-custodial wallet with instant zero-fee transactions, reaching more than a billion users, has the potential to be a breakthrough moment. But Durov's team left out almost everything that matters. Who will operate it? How will recovery work? Which networks will it support? The answers are still missing.
The direction Telegram's announcement signals matters more than the wallet itself: crypto infrastructure moving into the apps people already use.
Yana Mar, CBDO and Head of Account Management at ChangeNOW, has a straightforward take:
The implementation will be the deciding factor. It's hard to evaluate a product based solely on marketing.
A billion users is just a number until you know what they'll actually do with the product once it's in their hands.
The same logic applies further up the stack, where the participants aren't even human. AI agents are already spending money independently — buying data, computing power, and digital tools. Keyrock has tracked over 176 million such payments, mostly in USDC, often for pennies. In February 2026, developer Nik Pash asked an agent to send 4 SOL, and it sent 52.4 million LOBSTAR instead — about 5% of the token's total supply — costing roughly $450,000 simply because no spending limit had been set.
Major players like Amazon Bedrock AgentCore Payments and Cloudflare's Monetization Gateway are already embedding this AI agent payment infrastructure into their platforms. Machines paying machines without human intervention represents the next layer of adoption, happening quietly in the background.
Products now have to serve both human users and autonomous agents. ChangeNOW's AI Hub was built with that in mind, offering a single entry point for both audiences. Elias Viločkin, Chief Product Officer at ChangeNOW, describes the motivation:
We saw growing demand from our clients for specialized AI tools to work with crypto.
Elias adds that a significant share of traffic already comes from AI systems. Building for an AI-native environment goes beyond adding a chatbot. It requires structured documentation, machine-readable specifications, and dedicated integrations that automated systems can work with.
Culture matters too, often in ways that get dismissed as noise. Memes and online communities create entry points for new users, frequently before those users even think of themselves as "crypto people." In a recent X Space with leaders from BabyDoge, FirstBroccoli, and Turbo, the conversation turned to what makes some meme coins survive while most vanish within weeks. Participants pointed to community, consistency, and trust built through participation. Ekin, former Head of Content at BabyDoge, made a simple observation:
Meme culture is also Web 2.0. People share memes on Web 2.0.
What's already a cultural activity becomes an economic one through tokenization.
Look at the bigger picture and the pattern is clear: crypto continues a 10,000-year arc of how humans exchange value. Before money existed, there were debts, reputation, and records. Clay tokens from the ancient Near East were a system of accounting long before writing emerged, and trust was built on shared records rather than physical coins. Every step in this evolution — from temple storehouses to blockchain, from oral promises to smart contracts — has been about solving the same problem: how to exchange value with people you don't personally trust.
Crypto is becoming part of the background rather than a destination, and that shift changes the relationship between platforms and their users. Perhaps the next phase won't be about convincing people to try something new; it will be about delivering value so seamlessly that the technology itself becomes invisible.
Thanks for reading, and see you in the next edition.
