Session replay intelligence CLI

Make product analytics actionable for AI agents.

Run global quickstart once, initialize each repo with vesai init, then use replay evidence to generate user stories, group stories, and research answers your agents can execute. Commands auto-sync the CLI to latest main before running.

View on GitHub
Requiresgitbungcloudffmpeg
Terminal — vesai
$ curl -fsSL https://ves.ai/install | bash
Installing VES AI...
Linked vesai to ~/.local/bin/vesai
$ vesai quickstart --max-render-memory-mb 8192
Auto-updating VES AI from origin/main...
VES AI quickstart (global core setup)
Render budget set to 8192 MiB (dynamic scaling at 512 MiB/service)
Global quickstart complete.
$ vesai init --lookback-days 180
Created .vesai/project.json
Added .vesai/ to .gitignore
Project init complete.
$ vesai user bryce@company.com
Found 12 sessions for bryce@company.com
Rendering + analyzing sessions...
{
"email": "bryce@company.com",
"sessionCount": 12,
"userScore": 78,
"markdownPath": ".vesai/workspace/users/bryce-company-com-bryce-company-com.md"
}

How It Works

From machine setup to project intelligence artifacts in four steps.

Quickstart

Configure global runtime + render memory budget

Init

Create project-scoped .vesai config and workspace

Heartbeat

Daemon backfills and continuously analyzes new sessions

Ask

Use user, group, and research commands for decisions

See It in Action

Real commands, real output. Every data command returns structured JSON by default.

Terminal — vesai
$ vesai quickstart --max-render-memory-mb 8192
Auto-updating VES AI from origin/main...
VES AI quickstart (global core setup)
Configured core runtime at ~/.vesai/core.json
Render budget: 8192 MiB (dynamic service scaling enabled)
Global quickstart complete.
$ vesai init --lookback-days 180
VES AI init (project setup)
Created .vesai/project.json with UUID projectId
Created .vesai/workspace/{sessions,users,groups,research}
Added .vesai/ to project .gitignore
Project init complete.

Built for AI Agents

Same commands, same outputs whether you're a human or a coding agent.

Machine-Readable by Default

Every data command emits JSON. Pipe output directly into your agent workflows, CI pipelines, or dashboards.

Durable Workspace Artifacts

Session, user, group, and research artifacts persist as git-friendly markdown in .vesai/workspace/ for long-lived agent context.

Local-First, Self-Hosted

Global machine config stays in ~/.vesai while project credentials and artifacts stay in each repo's .vesai directory.

Ships with a SKILL.md

A comprehensive skill file teaches Claude Code, Codex, and other coding agents how to run quickstart/init plus user, group, and research workflows out of the box.

Integrations

PostHog
PostHog

Session recordings & product analytics

Google Cloud
Google Cloud

Vertex AI, Gemini, Cloud Storage

Playwright
Playwright

Browser-based replay rendering

ffmpeg

Video & frame processing

Getting Started

Prerequisites

  • gitVersion control
  • bunJavaScript runtime
  • gcloudGoogle Cloud SDK, authenticated
  • ffmpegVideo processing

Authenticate gcloud:

gcloud auth login
gcloud auth application-default login
gcloud config set project <project-id>

Four Steps

Every command auto-syncs VES AI to latest main before execution.

  1. 1Install VES AI
    curl -fsSL https://ves.ai/install | bash
  2. 2Set up global runtime
    vesai quickstart --max-render-memory-mb 8192
  3. 3Initialize this project
    vesai init --lookback-days 180
  4. 4Run replay intelligence
    vesai user <useremail>