Loading...
Features
Built for sales reps and individual sellers. The tool for the speaker, not the listener. Every feature, from pre-call prep to the coaching trajectory you build over time.
Before the call
Walk in prepared, not winging it
Upload your deck, set a goal, review an AI-generated Game Plan, scan briefing cards from past interactions, and skim a counter-objection library mined from session history. Cmd+Enter starts the call once 3 of 4 prep steps are green.
The moment you open a contact, Sussur drops in a short summary plus a “Watch for” list pulled from past sessions. No clicks, no waiting.
Personas evolve across 7 structured fields: communication style, personality notes, hot topics, trigger words, preferred approach, past outcomes, and raw context. Each call updates the relevant fields.
Add a contact and People Data Labs auto-fills role, company, LinkedIn, industry, seniority, headline, and Gravatar photo. The enrichment badge shows live state — Enriching, Enriched 5 days ago, From calendar, or No public profile match.
A personalized briefing card on the PrepView right rail, generated from your profile baselines and active insights. Reminds you of the techniques that worked last time with this kind of conversation.
Connects via OAuth and surfaces upcoming meetings on Home with attendee-to-contact matching and process-based detection for Zoom, Teams, FaceTime, and Meet.
During the call
Real-time intelligence while you talk
Four types of AI suggestion surface in real time: technique recommendations (blue), questions to ask (coral), warnings when something is going wrong (red), and affirmations when you’re on track (green). Each suggestion ships with a copyable “say next” snippet and a global keyboard shortcut to accept.
Capture · v0.10.25
An always-on-top coaching panel stays visible over Zoom, Meet, or Teams. Compact mode (50 px) shows a status bar with timer, coaching dot, and rapport sparkline. Extended mode (400 px) shows the full dashboard. Cmd+Shift+O toggles between them.
Capture · v0.10.25
Six linguistic signals scored every 30 seconds on a 5-minute sliding window — turn balance, response latency, backchannel frequency, question reciprocity, lexical convergence, and agreement signals. Composite 0-100 with EMA smoothing and inflection alerts on 15-point drops.
Capture · v0.10.25
Dual-color progress bar with proactive nudges when you dominate the conversation. WPM pace shown separately with zone labels.
Capture · v0.10.25
MediaPipe face mesh detects faces on screen and classifies expressions — smiling, frowning, surprised, confused, neutral. Expression shifts in the other person trigger coaching hints. Fully on-device, no cloud processing.
Capture · v0.10.25
Live topic detection plus open-question, tension, and decision tracking. Counts surface as chips in the context bar with full lists on hover.
Capture · v0.10.25
Type to the coaching AI mid-call. Quick prompt pills (“What should I say?”, “Help with objection”), slash commands /kb and /summary, up to 50 messages per session.
Capture · v0.10.25
3D-Speaker eres2net model labels speakers as they talk. Tap “Speaker 2” to rename it “Sarah” and the label sticks across calls via cross-call matching.
Capture · v0.10.25
Every 15 seconds, an LLM pass classifies the emotional register of the conversation. Surfaces in the context bar as a chip and as a sparkline in the extended overlay.
Capture · v0.10.25
Coaching can speak in your ear via macOS say or OpenAI TTS. Route to left ear, right ear, or both. A half-duplex playback gate keeps the TTS out of the transcript.
Capture · v0.10.25
Toggle the toolbar MagicWand and Sussur cleans transcript segments inline with a hallucination guard. Off by default.
Capture · v0.10.25
Cmd+K or one click bookmarks a moment. Marks show up as coral ticks on the rapport timeline and as cards in the post-call report.
Capture · v0.10.25
Optional WAV recording of mic and system audio. Configurable retention (default 30 days). Drives session replay scrubbing.
After the call
Every meeting becomes actionable intelligence
AI-generated dossier with summary, goal assessment, key moments, techniques used, action items with owners, next steps, relationship notes, and a draft follow-up email. Includes rapport radar, emotion timeline, coaching effectiveness, and optional comparison with the previous session.
Scrub past calls with time-synced transcript, screenshots, rapport timeline, emotion shading, and state snapshots. Face bounding boxes overlay screenshots. Audio playback drives the timeline. A post-session diarization pass corrects realtime speaker labels.
Rapport radar with all 6 signals, talk-time chart per speaker, conversation-flow timeline, and coaching effectiveness (suggestions fired vs accepted). Compare mode overlays previous-session data.
Capture · v0.10.25
Set a goal pre-call, get a graded assessment post-call. The model says whether the goal was met, partially met, or missed, with one-paragraph reasoning grounded in the transcript.
Capture · v0.10.25
Export reports as PDF, Markdown, JSON, HTML, or copy to clipboard. Session replay includes face detection data, expression distribution, and face-presence rates in every format.
Your coaching trajectory
Get measurably better, week over week
Eight anchor-nav sections — Identity, Trajectory, Mechanics, Skills, Goals, Playbook, Moments, People — that turn every past call into a self-improvement loop. Sparkline-driven mechanics grid, 7-dim skill radar, structured goals with follow-through detection, BestMoments, TopContacts, ActivityStrip.
A short Source Serif italic essay summarizes what changed this month — where rapport is trending, which patterns are setting in, what to work on. Capped at 4 generations per month with a 24-hour manual cooldown.
Seven coaching dimensions plotted as a radar with shrinkage applied — low-sample scores are pulled toward the baseline so noise doesn’t dominate.
Eight cards covering the mechanical levers of your call performance — talk ratio, question rate, pause discipline, and more. Each card has a sparkline, a delta-vs-baseline, and a benchmark band.
A self-learning user profile extracts patterns from your sessions — “You tend to over-explain when interrupted”, “Rapport drops faster than average when the other speaker uses jargon you don’t acknowledge.” Accept or dismiss each insight to tune the model.
Privacy & control
Your data, your rules
Four local STT models, your choice of size vs accuracy — Parakeet TDT 0.6B v3 (640 MB, default, highest accuracy), Whisper Tiny (75 MB, multilingual), Whisper Base (141 MB), Whisper Small (460 MB). Audio never leaves your Mac.
Claude, GPT, or Gemini via OpenRouter using your own API key. Per-subsystem model selection across coaching, report, summarizer, expression, emotion, and polish. Defaults to Gemini 3.5 Flash. No vendor lock-in, no markup.
Run coaching through Ollama for fully offline AI. Combined with on-device transcription, you get a complete coaching system that works without internet.
Track LLM spend across 8 period presets, KPI cards with delta vs previous period, a daily chart with anomaly markers and plain-language cause inference, forecast pacing with confidence band, per-model breakdown, top sessions, per-contact sparklines, monthly budgets and alerts, CSV export, prompt-version tracking, and a reconciliation surface. A typical coaching session costs $0.05-0.15.
Hide Sussur from screen sharing with one toggle. The window renders black in recordings and screen shares, keeping your coaching completely invisible to other participants.
RAG-powered retrieval across Cialdini’s persuasion principles, Chris Voss’s negotiation tactics, and rapport-science frameworks. Add your own documents. Hybrid vector + full-text search with LLM reranking. Knowledge graph view with entity extraction and community detection.
Single scrollable settings page with inline advanced disclosures, instant search, 4 built-in presets (Sales Discovery, One-on-One, Team Standup, Default), and one-click import/export.
Watch six linguistic signals build a live rapport metric during a real conversation.
[ Video placeholder ]
Six linguistic signals building a rapport metric during a real conversation. Video coming soon.