llms.txt Content
# Velt
> Embeddable review and approval for AI-native apps. Velt is an SDK that adds comments, approval workflows, AI review agents, suggestions, audit trails, memory, and notifications to your product. Agents do the work, humans decide.
Velt is built for products where work cannot ship unapproved: sales enablement, fintech and FP&A, compliance, operations, and AI-native SaaS. Agent suggestions become comments a human approves; approved changes fire through your webhook with a permanent record.
## Docs
- [Quickstart](https://velt.dev/docs/get-started/overview): install and render your first comment in minutes
- [API Reference](https://velt.dev/docs/api-reference): full SDK surface
- [Components](https://velt.dev/docs/components): React, Next.js, Angular SDK components
- [Docs (full)](https://velt.dev/docs/llms-full.txt): complete docs content as a single file
## Features
- [Comments](https://velt.dev/comments.md): contextual threads from humans or agents on any element
- [Approval flows](https://velt.dev/approval-flows.md): user-defined review pipelines with routing, conditions, quorum
- [Review agents](https://velt.dev/review-agents.md): AI first-pass review from plain English instructions
- [Suggestions](https://velt.dev/suggestions.md): propose and accept edits like a diff, from humans or agents
- [Audit trail](https://velt.dev/audit-trail.md): immutable, exportable record of every action
- [Memory](https://velt.dev/memory.md): past decisions surface as precedent
- [Notifications](https://velt.dev/notifications.md): in-app, email, Slack, Teams
- [Presence](https://velt.dev/presence.md): avatars, cursors, live selection, and follow mode for humans and agents
- [Multiplayer editing](https://velt.dev/multiplayer-editing.md): co-editing, single editor mode, and live state sync
- [Recording](https://velt.dev/recording.md): voice, video, screen, and a built-in video editor
- [Huddle](https://velt.dev/huddle.md): live audio and video inside the document
- [Self-ho