Launching Fall 2026
Acoustic intelligence for
Amazon Connect
Always-on silence detection, talk patterns, and real-time call visibility — delivered through your AI agent. No transcription. No recordings stored.
~$0.003/min · 1,000 minutes free every month · No credit card required
What Jaypaper does
AI-like analytics, not AI-like prices.
Acoustic signal, not words
Voice activity detection on raw audio. We measure speech presence, silence, crosstalk, and timing patterns — never the content of a conversation.
No recordings stored
Audio is analysed in real-time and discarded. No transcripts, no recordings, no PII in storage. The meaning of the call is never captured.
Delivered through your AI
Jaypaper is an MCP server. Ask Claude about your calls and get answers with inline charts and tables. No separate dashboard to learn.
Every call, all the time
Always-on monitoring across every call — not just flagged ones. Real-time silence detection, crosstalk alerts, and talk pattern analysis.
5-minute setup
Deploy one CloudFormation template. Add two blocks to your contact flow. Connect your AI. That's it — no agent desktop changes.
8× cheaper
~$0.003/min vs Contact Lens at $0.015–0.025/min. We don't transcribe, so we don't pay transcription costs. You save the difference.
How it works
Deploy our template
Run our CloudFormation template in your AWS account. Creates a Kinesis stream and grants Jaypaper read-only access. ~5 minutes.
Jaypaper processes your calls
Acoustic signals are analysed in real-time — speech presence, silence, crosstalk. No recordings leave your account.
Ask your AI
Connect Jaypaper to Claude (or any MCP-compatible AI). Ask about your calls — get answers with charts and tables rendered inline.
Architecture
Your data stays in your AWS account. Jaypaper only reads the acoustic signal.
Your AWS Account
Amazon Connect
↓ audio streams
Kinesis Data Stream
~$11/mo
Jaypaper
Voice Activity Detection
↓
Real-time Analytics
↓
MCP Server
Your AI
Claude / MCP Client
“How are my calls?”
→ charts, tables, insights
Privacy by architecture
Jaypaper processes acoustic features only — speech presence, energy levels, timing patterns. No recordings are stored. No transcripts are generated. No words are captured.
- ✓The “meaning” of the call is never stored
- ✓No PII in storage — no compliance review of recordings
- ✓Customer owns their Kinesis stream — revoke access anytime
- ✓Less liability, less exposure, same operational insight
Stay in the loop
We're launching fall 2026. Get notified when early access opens — no spam, just launch updates.
Powered by Buttondown. Unsubscribe anytime.