Descriptions:
Sharbel A. tackles a common OpenClaw productivity problem: when all conversations flow through a single Telegram chat, the AI agent loses context separation, conflating unrelated threads — trading ideas bleeding into YouTube script drafts bleeding into client requests. The solution is Telegram’s built-in Topics feature, which OpenClaw treats as entirely separate conversation contexts with independent memory.
The video walks through the full four-step setup: creating an OpenClaw bot via BotFather and saving the token; creating a Telegram group with Topics enabled and adding the bot as administrator; configuring BotFather settings (allow groups on, group privacy off so the bot responds without being @-mentioned); and adding the group ID to OpenClaw’s configuration. The config edit can be done manually in the JSON file or by simply asking the OpenClaw agent itself to handle the setup given a group ID — the easier path the host recommends.
Sharbel shows his own topic structure as a template: dedicated channels for YouTube, X/Twitter content, trading (including Polymarket and Hyperliquid), active builds, a site-specific topic, and a general catch-all. Each topic maintains its own conversational memory, effectively giving one AI agent multiple specialized personas across different life and work domains. The pattern is directly applicable to any OpenClaw deployment using Telegram as its interface.
📺 Source: Sharbel A. · Published April 10, 2026
🏷️ Format: Tutorial Demo
