Direct answer
What is Vellum?
Vellum is an autonomous personal AI assistant that executes real tasks across computer environments, apps, and channels using persistent memory and dedicated compute.
Vellum operates as a personal intelligence assistant running on a dedicated cloud machine or self-hosted hardware. Unlike conventional chat interfaces that reset context between sessions, Vellum maintains a persistent workspace filesystem and long-term memory system that stores facts, preferences, and operational context over time. The system is equipped with out-of-the-box skills covering email management, calendar coordination, web browsing, voice calls, code execution, web application creation, and cross-channel messaging across Slack, Telegram, web, terminal, and mobile interfaces. Actions execute within an isolated workspace, and operations that interact with external services, shell commands, or screen control require explicit user-approved permissions or defined standing trust rules. Vellum uses an un-marked-up model pass-through pricing model, where token usage and credit-eligible features (like web search, image creation, and LLM inference) draw from monthly plan allowances or prepaid credit balances. For developers and teams with data governance requirements, Vellum can also be self-hosted locally or deployed onto private cloud infrastructure.
Vellum combines persistent memory, its own dedicated compute and workspace filesystem, and an explicit permissions engine that executes real-world actions across communications and development channels without charging markups on underlying LLM tokens.
Product capabilities
Key features
Persistent Memory Subsystem
Consolidates and extracts facts, preferences, and entities across conversations into a personal knowledge base, preserving context over time.
Multi-Channel Accessibility
Interact with the assistant across macOS, iOS, web browser, terminal, Slack, Telegram, email, and voice calls.
Dedicated Workspace and Computer
Runs on isolated virtual compute with configurable CPU and RAM tiers (from 1 vCPU up to 4 vCPU / 16 GiB RAM) and persistent storage.
Built-In Skill Library
Includes integrations for Gmail, Google Calendar, Slack, Linear, DoorDash, web search, phone calls, and interactive web app generation.
Granular Permissions Model
Prompts users with Allow or Deny options and risk badges before executing file system changes, shell commands, or outbound messages.
Self-Hosting and Extensibility
Open-source core architecture deployable locally or to private clouds like GCP, with support for custom skills, MCP, and hooks.
Workflow
How Vellum works
- Sign up or deploy the open-source assistant locally or to your own cloud infrastructure.
- Configure your assistant's identity, communication channels (Slack, Telegram, CLI, email), and preferred integrations.
- Delegate tasks conversationally or set up background routines and heartbeats for continuous work.
- Review and approve external actions through the built-in risk and permissions engine.
- Monitor resource utilization across persistent storage, compute machine tiers, and zero-markup token credits.
Practical fit
Who should use Vellum?
Daily Briefing and Schedule Management
Aggregate calendar events, unread messages, weather, and news into morning summaries, and make phone calls or update calendar invites.
Autonomous Inbox and Ticket Triage
Read and draft email replies, move tickets inside Linear, and push team updates into specific Slack channels.
On-Demand Micro-App Generation
Generate functional web utilities (such as habit trackers) directly within the conversation window.
Asynchronous Background Research
Trigger automated heartbeat routines to price-watch flights, assemble competitor briefs, and compile standup notes while away.
Editorial assessment
Pros and limitations
Where it is strong
- Zero token markup on LLM provider costs
- Persistent memory that carries across conversations and devices
- Open-source core with local and private cloud self-hosting options
- Multi-channel support including CLI, web, Slack, and Telegram
- Interactive action execution beyond basic text generation
Where to be careful
- Background processes like memory compaction and heartbeats consume usage credits
- Requires managing compute machine tiers, storage sizes, and credit balances
- Action execution requires careful configuration of security permissions and risk rules
Commercial context
Vellum pricing
At the review date (September 2026), Vellum provides a free Base plan (1 vCPU, 2 GiB RAM, 6 GiB storage) with pay-as-you-go credit usage. Pro plans start at $30/month (Mighty) up to $200/month (Ultra) with bundled machine capacity, storage, and usage. Self-hosting via GitHub is also available. Always verify current pricing on the official site.
| Plan | Price | What it includes |
|---|---|---|
| Base | $0 / forever |
Free forever |
| Mighty | $30 / month |
Does not include platform fee, custom subdomain, or static IP |
| Super | $100 / month |
Platform fee included |
| Ultra | $200 / month |
Platform fee included |
Pricing, limits, taxes, model access, and regional availability can change. Verify the purchase-critical details on the official pricing page linked under Sources.
Transparent ranking
Why Vellum scores 70.7
Each factor is scored on a 100-point scale, then combined using the public ToolsRank weights. Engagement and momentum stay at a neutral baseline until measured signals exist, so no tool can gain or lose position from numbers nobody recorded.
Compatibility
Languages, platforms, and integrations
Languages
- English
Platforms
- Web
- macOS
- iOS
- Linux / Self-hosted
- Terminal
Integrations & surfaces
- Slack
- Telegram
- Gmail
- Google Calendar
- Linear
- Tavily
- Perplexity
- Brave Search
- Firecrawl
- DoorDash
- Stripe
Community
Reviews and questions
No approved member reviews yet. Editorial factors above are the only rating on this page.
Reviews and questions come from Google-signed members and are checked by an editor before they appear.
Frequently asked
Vellum FAQ
What is Vellum?+
Vellum is an autonomous personal AI assistant that lives on your computer or in the Vellum Cloud. It possesses its own identity, persistent memory, and dedicated filesystem, allowing it to send emails, manage calendars, browse the web, execute code, and perform actions across integrated services.
How is Vellum different from ChatGPT or Claude?+
Unlike standard chatbots that operate as isolated text generation sessions, Vellum operates with persistent memory, an individual filesystem, and proactive routines (such as heartbeats). It can take real actions in external systems across terminal, mobile, web, email, and messaging platforms.
How does Vellum handle pricing and token usage?+
Vellum passes through model provider token costs at cost with no markup. Work draws from monthly included usage allowances or prepaid Vellum Credits (1 credit = $1 USD). Paid plans scale computer size (vCPU/RAM), storage, and usage bundles.
Can Vellum execute tasks without human confirmation?+
Only what you explicitly permit. Actions that touch the filesystem outside the workspace, control the screen, or send external emails require an Allow or Deny confirmation unless you configure standing trust rules.
Can I self-host Vellum?+
Yes. Vellum assistant is open source and can be deployed locally on your own machine, a VPS, or a private cloud like Google Cloud Platform (GCP). Self-hosting incurs no platform fee or computer tier charges.
