Telegram’s AI agent infrastructure is official, and the adoption is already here.
On May 7, Telegram rolled out its massive “AI Bot Revolution” update. The release introduced more than 10 new native features designed to turn the messaging giant into an artificial-intelligence operating system. From Guest AI Bots that can spin up answers via simple @username tags to autonomous Bot-to-Bot Communication modes, Telegram effectively built the full agent runtime environment: identity, presence, autonomy, and real-time streaming expression.
If Telegram just finished laying the plumbing, The Open Platform (TOP) is showing everyone the water.
An App People Can Use
Today, TOP announced the global launch of Mira, a messenger-first AI agent built natively into Telegram. While standalone AI applications have spent the last year demanding that users context-switch, jump into separate browsers, or duplicate workflows, Mira lives precisely where digital coordination already happens: inside group chats and personal conversations.
The market response proves that consumers were waiting for exactly this approach. Moving well past a simple engineering demo, Mira has already crossed 2 million users. The product is actively adding value in more than 50,000 Telegram groups, with monthly active users crossing the 500,000 mark and doubling month-over-month.
This is a textbook case of a consumer application validating an ecosystem layout. When Telegram rolled out Guest Bots to let unintegrated agents reply to mentions, it validated the collaborative-group context that Mira pioneered during its quiet phase. When the platform announced bot-to-bot infrastructure to automate cross-agent workflows, it was chasing a vision Mira already delivers daily through more than 900 service integrations, including Google Calendar, Notion, Gmail, GitHub, and Canva.
Distribution Strategy Wins the Wave
The broader story unfolding here is less about isolated technical features and entirely about the mechanics of AI distribution. Telegram surpassed 1 billion monthly active users in March 2025, handing native applications a massive adoption pipeline that standalone assistants simply cannot replicate.
More than a third of Mira’s new users discover the agent dynamically by watching it execute tasks inside shared chats before adopting it for their own personal workflows. It is an organic, viral acquisition loop that bypasses the friction of the traditional app-store download process.
Once user demand is triggered, there is no technical friction. Inside Telegram, users can add @mira to a group or message it directly without any additional setup. Once added, Mira handles a broad spectrum of daily tasks, turning shared chats into spaces where people can coordinate, execute tasks, and interact with AI together.
Users are utilizing it for everything from project management and trend spotting to generating promotional AI video assets and tracking group schedules. Its most highly trafficked specialized agents focus on financial insights, career planning, and nutrition guidance.
Behind the scenes, Mira utilizes a dynamic routing architecture. Rather than locking users into a single model, it routes individual tasks to major external providers such as OpenAI, Anthropic, Minimax, and ElevenLabs to balance execution speed with response quality.
“People don’t want to interact with Al in isolation,” Daria Yakovleva, CEO at Mira, said. “They want an agent that works with them where decisions are already made, inside personal and group chats. Mira is turning chat into a single interface for the internet, where people can search, plan, and act without ever leaving the conversation.”
Privacy via Blockchain Infrastructure
For enterprise teams or users demanding tighter data perimeters, Mira includes a Private Mode powered by Cocoon. Cocoon operates as a decentralized GPU network developed by Telegram and anchored to the TON blockchain. Turning on Private Mode allows data requests to bypass external corporate cloud providers entirely, routing the processing across a decentralized, private cryptographic infrastructure instead.
Looking ahead, TOP plans to expand Mira’s native footprint by leveraging the platform’s newly available infrastructure. The roadmap highlights agent-to-agent interactions and the deployment of a dedicated payment sub-wallet. This sub-wallet will allow users to securely delegate transaction authority to the agent, enabling Mira to purchase digital services, book tools, and complete automated checkouts within predefined financial limits.
The era of navigating isolated, fragmented AI web tabs is coming to an end. The next wave of adoption belongs to agents that meet humans inside their existing social structures. The infrastructure is live, the playbook is clear, and Mira is emerging as the definitive proof that the Telegram agent economy has officially arrived.
