When you return to ZapRag after time away — a 2-hour lunch, a weekend trip, a long meeting, an entire vacation — Radar shows an intelligent summary of what appeared while you were busy. Instead of listing every message, it identifies and prioritizes what really needs attention: pending items waiting for your reply, questions directed at you, decisions affecting your work, and direction changes in projects. Focus on the signal, discards the noise. That group chat with 200 messages becomes 3 lines of 'what matters to you'. Dramatically reduces re-entry time after any pause.
WhatsApp and Telegram, individual conversations and groups, across all connected sessions. Radar consolidates everything in a single panel organized by conversation, with visual priority badges (urgent, attention, informational). Smart direct mention detection: recognizes when someone called you by name (even variations like 'Adriano', 'Adri', 'AP') or replied to your message, even in groups with many people talking at once. Supports detection in English, Portuguese, and Spanish. Ignores generic mentions like 'everyone' or 'team' (unless you flag that conversation as critical).
You can trigger Radar manually anytime (button at the top of the panel, or 'radar' text command), or leave it active to appear automatically when you open the platform after a configurable inactivity period (1h, 4h, day, weekend). Updated in real time: as new messages arrive, Radar can be refined incrementally, no need to reprocess everything. Radar history available for 30 days — useful for reviewing 'what happened during my absence last week' or 'what did I follow up with that client in October'. Optional scheduled trigger: receive via WhatsApp every day at 9am.
Radar uses Insight as a generation engine, but with a specialized prompt for executive synthesis. When you open the platform after a period of inactivity, the system identifies all messages that arrived in that interval, groups by conversation, and passes contextualized batches to the LLM (Claude Sonnet by default) with specific instructions: identify pending items, directed questions, decisions, status changes. Different weights per conversation: conversations marked as critical get higher priority, silenced conversations only appear if there's a direct mention. Result is presented in descending order of calculated relevance.
Professionals returning from vacation who want to understand in 5 minutes what changed over 15 days of accumulated WhatsApp. Parents who check school groups once a day and need to know if there was an important note. Salespeople returning from travel who want to see pending proposals without reading 800 messages. Managers covering for another manager's vacation who need to absorb context quickly. On-call duty (medical, legal, technical): whoever takes a shift sees a summary of the previous shift before starting. Night-shift entrepreneurs who check what happened at the company in the morning without losing an hour in chat. Anyone accumulating 200+ unread daily.
Radar pulls data from Sessions (all connected conversations), uses Insight as AI engine, can dispatch conversations to Tasks (if it identifies something needs to become a task, with 1 click it becomes a Kanban card), and its summary also enters Reporter's briefing (configurable). You can mark a specific conversation as 'critical' (always appears in Radar even without direct mention) or 'silenced' (only appears if you're mentioned by name). Additional filters: only client conversations, only groups, only individuals, only messages containing money/deadline/people.
Q: Does Radar read private messages? — Yes, because RAG already indexed everything. Synthesis happens locally on ZapRag, your messages never leave our server for third parties beyond processing purposes. Q: Does it work in groups with 500 people? — Yes, and that's where it helps most. Q: Can I receive Radar via WhatsApp? — Yes, configure a time and ZapRag itself sends it to you. Q: How much does each Radar cost? — Included in all plans with no execution limit. Q: Does Radar summarize media (audio/images)? — Audio yes (transcribed via Whisper), images with text yes (via OCR), videos partially (audio only extracted).