Previous Case Study • Mobile / SMB
BizNest
Unified workspace app consolidating invoicing, chats, and scheduling.
Case BNRead full study →
Back to projects
AI / Productivity
InboxFM
AI-native email assistant designed for priority-driven workflows.
96.4%
Sort Accuracy
45m/Day
Time Saved
< 80ms
Local Latency
inboxfm.velnix.studio
InboxFM Client(3 unread)Vim mode
Aangi Shah (Design Studio)ACTION NEEDED
Review layout templates for team details
"We need to approve the final colors by Friday."
AI THREAD SUMMARY (3 EMAILS)
The client approved the Indian Rupee pricing options (Lakhs). Staggered transition changes for the sidebar menu are complete. Design system components are ready to test.
Overview
InboxFM uses local embedding models running directly in the browser to organize and summarize inbound emails. It prioritizes correspondence requiring actions and summarizes complex threads, saving professionals hours of inbox sorting every day.
The Stack
React / ViteTypeScriptTransformers.jsRedisNode.js IMAPPostgreSQL
What We Shipped
Transformers.js local embeddings model integration
Thread summary pipeline
IMAP/SMTP socket synchronization service
Vim-inspired keyboard navigation layouts
The Challenge
Modern email clients are cluttered with notifications, marketing materials, and long threads. Finding emails that actually need a reply is tedious and leads to delayed business decisions.
Our Solution
We created a local first email client. Integrating lightweight client-side transformers, InboxFM scans inbox deltas, highlights emails containing active questions or requests, and compiles thread updates in key bullet points.
Keep Exploring
Other Case Studies
Next Case Study • AI / Documentation
Doxify
Outdated-page scanner and automated team QA assistant.
Case DXRead full study →