# Virtual

> Virtual is an AI-powered video intelligence layer for creators. It turns screen recordings into interactive knowledge bases that answer viewer questions, surface confusion via per-second heatmaps, and cluster similar questions into "Master Doubts" that a creator can answer once for every future viewer.

## Core concepts

- **Ask Assistant**: An AI-powered Q&A panel that floats on the right side of any recorded video. Viewers can pause at any timestamp and ask any question. The AI answers using the video's transcript, frame analysis, and the creator's saved ground-truth answers.
- **Confusion heatmap**: A per-segment density map showing where in a recording viewers asked the most questions. Color-coded: gray (no questions), green (1–2), yellow (3–5), orange (6–9), red (10+).
- **Master Doubts**: Semantic clustering of similar viewer questions into a single canonical phrasing with a count. A creator answers once; the AI handles every future viewer who asks something similar.
- **Ground-truth training**: The creator types an answer in their own words. The AI uses that exact phrasing — not a generic LLM response — for matching future questions.
- **Mental State Tracking**: Automatic classification of viewer messages as either "content doubts" (real questions about the video) or "conversational" (greetings, small talk, AI-identity questions). Only content doubts surface to the creator; conversational messages are tracked but filtered out.
- **Wake word**: Saying "Hey Virtual" while watching a video opens the Ask Assistant panel hands-free.

## Pricing

Virtual is priced by viewer questions per month, not seats. The cost driver is also the value driver.

- **7-day free trial**: 100 viewer questions, 10-min max video, 500 MB storage, up to 720p source files
- **Basic — $19/mo**: 600 questions/mo, 20-min max video, 8 GB storage, up to 720p source files, Studio editing
- **Pro — $49/mo**: 1,800 questions/mo, 45-min max video, 30 GB storage, up to 1080p source files, priority processing
- **Premium — $119/mo**: 4,500 questions/mo, 90-min max video, 100 GB storage, up to 4K source files, priority support
- **Enterprise**: Custom; SSO, on-prem storage, SOC 2, DPA, dedicated SLA

## Who uses Virtual

- Tutorial creators publishing courses on course platforms, bootcamps, and private communities
- SaaS teams running customer onboarding flows
- Internal L&D teams replacing all-hands Q&A with async training
- Sales teams sending async demos to prospects
- Product walkthrough creators (PMM)
- Developer-relations teams shipping code walkthroughs
- Online educators running cohort-based courses
- Agencies delivering recorded client briefs and audits

## How Virtual differs from related tools

- **vs Loom**: Loom records and shares. Virtual makes the recording interactive — viewers ask, AI answers, heatmap reveals confusion. Loom shows who watched; Virtual shows where they got lost.
- **vs Vidyard**: Vidyard is built for sales outbound with deep CRM hooks. Virtual is built for letting many viewers interrogate the same recording.
- **vs Wistia**: Wistia is the premier video host with lead-gen forms and brand-customizable players. Virtual is a comprehension layer, not a host.
- **vs Descript**: Descript is creator-side AI (text-based editing, voice cloning). Virtual is audience-side AI (Q&A, heatmap, clustering).
- **vs Tella**: Tella records gorgeous video in the browser. Virtual takes any final MP4 and adds the interactive layer.
- **vs Trupeer / Guidde**: Those generate demos automatically from screen clicks. Virtual takes finished videos and adds Q&A on top.
- **vs Storylane**: Storylane builds clickable product simulations. Virtual makes recorded videos answer questions.
- **vs Riverside**: Riverside is studio-quality remote recording. Virtual is interactive playback.
- **vs Tango**: Tango captures click-by-click written guides. Virtual adds Q&A to video walkthroughs.

## Technical details

- Upload format: MP4 only, up to 10 GB
- Transcription: Whisper-class, embedded for semantic search
- Frame analysis: GPT-4o-mini vision, cached descriptions
- Answer LLM: Claude Sonnet for streaming SSE responses
- Storage: S3-backed, presigned URLs for playback
- Auth: Supabase
- Pipeline: BullMQ + Redis (transcode → keyframes → transcribe → embed → cleanup)

## Pages

- [Home](https://heyvirtual.me/): Product overview, features, pricing, FAQ
- [Blog](https://heyvirtual.me/blog): Notes on building Virtual
- Solutions: [Tutorial creators](https://heyvirtual.me/solutions/tutorial-creators), [SaaS onboarding](https://heyvirtual.me/solutions/saas-onboarding), [Internal training](https://heyvirtual.me/solutions/internal-training), [Sales demos](https://heyvirtual.me/solutions/sales-demos), [Product walkthroughs](https://heyvirtual.me/solutions/product-walkthroughs), [Developer docs](https://heyvirtual.me/solutions/developer-documentation), [Online educators](https://heyvirtual.me/solutions/online-educators), [Async standups](https://heyvirtual.me/solutions/async-standups), [Customer education](https://heyvirtual.me/solutions/customer-education), [Agencies](https://heyvirtual.me/solutions/agency-deliverables)
- Comparisons: [Loom](https://heyvirtual.me/compare/loom), [Vidyard](https://heyvirtual.me/compare/vidyard), [Wistia](https://heyvirtual.me/compare/wistia), [Tella](https://heyvirtual.me/compare/tella), [Descript](https://heyvirtual.me/compare/descript), [Trupeer](https://heyvirtual.me/compare/trupeer), [Storylane](https://heyvirtual.me/compare/storylane), [Riverside](https://heyvirtual.me/compare/riverside), [Guidde](https://heyvirtual.me/compare/guidde), [Tango](https://heyvirtual.me/compare/tango)

## Contact

- Email: team@heyvirtual.me
- Sales: team@heyvirtual.me