Get Free API Key

INTEGRATIONS

Add Velt to the editors, grids, and canvases you already use.

Comments, co-editing, presence, and review, anchored to the surface and stable as it changes, for your users and your AI agents. Agents suggest on the surface, a human approves, and the core library reaches anything that renders in a browser.

Get Free API KeyBook Demo

Free tier. No credit card. First comment in 5 minutes.

TIbrief-q3.md / Tiptap
live
BIUH1H2</>

The Q3 campaign centers on the migration storyHope: moving from async reviews to live collaborationEthan. Pricing: three usage tiers from $0, no per-seat cost.

Open
MayaMaya14m

Let's lead with this. Strongest angle we have, @Ethan can you tighten the wording?

live cursors + a comment anchored to the selection · Yjs merges every keystroke

What it is

One surface, humans and agents, a human approves.

Velt attaches its review primitives to the surface you already render, then puts your users and your AI agents on it together: an agent comments or suggests on the element, and a human approves or rejects before anything changes. Comments, presence, cursors, suggestions, and co-editing anchor to the element a user is pointing at, not a Velt-owned canvas. You add Velt three ways: a native adapter for deep integrations like co-editing, the custom-component API for comments and suggestions on any element, and the framework-agnostic core library or web components for surfaces with no adapter.

Agents as first-class users

Your AI agent comments and suggests on the surface as a first-class user; a human approves or rejects, and on approve the change fires through your webhook with a permanent record.

Learn more

Comments

Threaded comments on the content, with replies, @mentions, reactions, and attachments, that survive edits.

Learn more

Co-editing

Conflict-free multiplayer editing, so two people work the same surface without overwrites.

Learn more

Annotations

Shared, live-synced markup on a region, shape, or data point for read surfaces like PDFs and charts.

Learn more

Presence and cursors

Live named cursors, selections, and who is here right now.

Learn more

Suggestions

Propose changes as accept-or-reject suggestions, reviewed like a diff.

Learn more

Notifications

In-app, email, and Slack alerts on every comment and @mention, so a review never stalls.

Learn more

Recording

Pin a quick voice, screen, or video clip in context when showing beats typing.

Learn more

Version history

Named checkpoints, saved and restored by API.

Learn more

How it works

How it works

01INSTALL
terminal
npm install @veltdev/react
02PROVIDER
App.tsx
import { VeltProvider } from "@veltdev/react";

<VeltProvider apiKey={API_KEY}>
  <App />
</VeltProvider>
03MOUNT
Editor.tsx
// Mount the Velt primitive on your surface
<VeltComments />

Skip the steps. Have your agent set it up.

An in-house version of multi-surface review requires a per-editor binding for every library you support, anchoring that survives re-renders and concurrent edits, selection and cursor mapping per surface, a custom-element path for grids and canvases, and a re-test of all of it on every upstream editor release. Teams build this one editor at a time and re-litigate it on every dependency bump. Velt carries the adapter set as packages and the long tail as the core library.

Integrations

Drops into the stack you already have.

Every logo links to its setup guide. Toggle a surface to see the exact snippet that mounts Velt on it.

Every surface we plug into.

Text and code editors

TiptapLexicalPlateJSCodeMirrorAce EditorQuillDraft.jsBlockNoteMonacoProseMirror

Documents, PDF and office

SuperDoc

Spreadsheets, grids and tables

AG GridTanStack Table

Diagrams and canvas

React FlowKonva

Charts and data viz

Chart.jsHighchartsNivo

Build with Velt

These are dev-time tooling, not surfaces, so they are not in the matrix. Cursor, Claude Code. Each bundles the Velt SDK as skills, rules, an expert agent, and an MCP server, so your coding agent scaffolds and wires Velt for you.

CursorClaude Code

Agents inside Velt

These are runtime agent connections, not surfaces. Chat SDK Adapter (build cross-platform bots that read and reply in your comment threads), MCP server (let agents read your review data and propose changes, with a human approving).

Chat SDK AdapterMCP serverbeta

Works with the rest of your stack

Frameworks

ReactNext.jsVueAngularPlain HTML

What works on each surface today.

SurfaceCommentsCo-editingSuggestionsPresenceAgents
TiptapYesYesYesYesYes
LexicalYesYesYesYesYes
SlateYesYesYesYesYes
PlateJSYesYesYesYesYes
CodeMirrorYesYesYesYesYes
Ace EditorYesYesYesYesYes
QuillYesYesYesYesYes
Draft.jsYesYesYesYesYes
BlockNoteYesYesYesYesYes
TinyMCEYesYesYesYesYes
MonacoYesYesYesYesYes
ProseMirrorYesYesYesYesYes
CKEditorYesYesYesYesYes
ApryseYesAnnotationsYesYesYes
NutrientYesAnnotationsYesYesYes
SuperDocYesYesYesYesYes
SpreadJSYesYesYesYesYes
AG GridYesYesYesYesYes
TanStack TableYesYesYesYesYes
React FlowYesYesn/aYesYes
KonvaYesYesn/aYesYes
Chart.jsYesAnnotationsn/aYesYes
HighchartsYesAnnotationsn/aYesYes
NivoYesAnnotationsn/aYesYes

Comments and agent review reach every surface. Co-editing is conflict-free multiplayer editing on editable surfaces; charts and PDFs carry annotations, the same shared state applied to a rendered surface. Agent review is available wherever comments are. If your surface is not here, the core library still anchors Velt to it.

No adapter? Velt still anchors to it.

If it renders in a browser, Velt attaches to it. The custom-component API puts comments and suggestions on any element with a stable target; the framework-agnostic core library (text, map, array, xml stores) wires a custom editor, whiteboard, or grid onto the same realtime engine; web components render the Velt UI outside React. This is the path for a proprietary editor or an in-house canvas, and it is why the grid is a starting set, not a ceiling.

In production

In production and where it fits

Legal

Clause comments and redline suggestions in a contract editor and in PDF; an agent flags a missing term, counsel approves.

See the use case

Fintech and FP&A

Cell-level comments and co-editing on a spreadsheet or grid; an agent flags an outlier cell, your reviewer approves.

See the use case

Sales enablement

Comments and approvals on decks, emails, and landing pages built in a rich-text or custom builder.

See the use case

AI-native SaaS

Agent findings as comments on a generated draft inside the product's own editor, approved before they change user work.

See the use case

Related

Related primitives

Comments

The primitive that reaches every surface.

Learn more

Multiplayer editing

Co-editing on editable surfaces.

Learn more

Presence

Cursors and selection on the surface.

Learn more

Enterprise

Enterprise-grade security and data control

Comment and review content on any surface can stay on your infrastructure via per-feature data providers, so adding Velt to an editor does not move that editor's content off your stack.

FAQ

Frequently asked questions

Which surfaces support co-editing, and which only comments?

Co-editing, conflict-free multiplayer editing, is available on every editable surface: the text and code editors, the grids, the canvases, and the docx editor. Charts and PDF viewers are read surfaces, so they carry shared annotations rather than co-editing. Comments and agent review reach every surface in the grid. The capability matrix above states it per surface.

My editor or grid is not listed. Can I still use Velt?

Yes. The custom-component API anchors comments and suggestions to any element, and the framework-agnostic core library wires a custom editor or canvas onto the same realtime engine. A missing logo means no prebuilt adapter, not no support.

How do I add comments to Monaco, AG Grid, or a PDF?

Install @veltdev/react, wrap your app in VeltProvider, and mount the comment primitive on the surface: an adapter where one exists, or the custom-component target on the element. The per-surface setup guide has the snippet.

Do AI agents work inside every integration?

Yes. An agent anchors to a surface the same way a human does: it posts a comment or a suggestion on the element, and a human approves or rejects before anything changes. Where comments work, agent review works.

Is this like Liveblocks for any editor?

Liveblocks is realtime collaboration focused on engagement: sync and presence, with agents acting on app state. Velt is review and approval: agents comment and suggest on the surface, and a human approves before anything changes. Same multiplayer layer, plus the consent step and the permanent record.

Can a bot or AI agent reply inside comment threads directly?

Yes, through the Chat SDK Adapter, Velt's vendor-official adapter for Vercel's Chat SDK. A webhook fires when a user @-mentions the bot or reacts, and the bot replies in that same thread, with the same bot logic running on Velt, Slack, Discord, and other Chat SDK adapters.

Do integrations cost extra?

No. Velt is priced on usage, not seats or integrations: you pay for documents with review activity in a month, with a free tier for development and early production. Every integration is part of the SDK.

Add Velt to the editors, grids, and canvases you already use.

Get Free API KeyBook Demo

Free tier. No credit card. First comment in 5 minutes.