Vellum

What is Vellum?

Vellum is a personal AI assistant designed to retain context and act across the tools a person uses. A request can be combined with relevant memories, files, connected services, and browser or computer tools before Vellum responds or performs an action.

This makes Vellum different from a conventional chatbot. A general-purpose assistant such as ChatGPT may primarily be used for conversation, drafting, analysis, or question answering. Vellum is oriented more toward an ongoing assistant relationship in which the system can remember preferences, access authorized information, follow workflows, and take actions with user approval.

What Vellum can do

Maintain memory and context

Vellum can store information such as preferences, project details, follow-ups, and other context for later conversations. Persistent memory is central to the product: the goal is to reduce the need to repeatedly explain the same background when returning to a task or project.

Research the web and work with files

Users can ask Vellum to search the web, retrieve information, navigate pages, and summarize findings. It can also read workspace files and create or edit longer documents. These capabilities support practical research and information-retrieval tasks, such as finding details in personal notes, reviewing documents, or assembling an initial research brief.

Connect to communication and productivity services

Vellum supports connections and channels including Gmail, Google Calendar, Slack, Telegram, email, and phone calls through documented integrations. Depending on the connection and permissions, it can summarize activity, help manage messages, track follow-ups, and assist with scheduling or calendar-related work.

Automate recurring work

The assistant includes scheduled tasks, reusable workflows, playbooks, skills, plugins, and support for MCP servers. These features allow users to turn repeated instructions into more consistent processes. Vellum also provides developer-oriented extension surfaces for custom tools and assistant behavior, although using them may require more technical configuration than ordinary chat.

Control supported computers

On supported desktop installations, Vellum can click, type, navigate applications, and interact with web pages. Computer control is permission-based: sensitive actions such as sending messages, making purchases, or accessing new services normally require approval unless the user creates a more permissive trust rule.

Use voice and media features

Vellum supports voice input, speech synthesis, transcription, image creation, and supported media-processing workflows. These features complement the assistant's main role; Vellum is not primarily an image, video, or voice-production application.

How people realistically use Vellum

Vellum is most useful when a task involves several pieces of personal context or more than one application. For example, a user might ask it to review recent messages and calendar commitments, identify follow-ups, research an issue on the web, and prepare a document. It can also help plan trips, meals, workouts, or schedules; locate information in files; draft and manage email; transcribe recordings; build small applications; and handle recurring workflows.

The practical advantage is continuity. Instead of treating every request as an isolated prompt, Vellum can use stored context and authorized tools to carry work forward. The trade-off is that users must configure connections and decide which actions the assistant may perform automatically.

Platforms and workflow access

Vellum is available through a browser-based application, macOS and Windows desktop clients, and mobile applications for iPhone, iPad, and Android. A Chrome extension supports browser automation. Linux users can use the web application and browser extension, but there is no shipped Linux desktop client according to the documented availability information.

Users can also interact with Vellum through connected channels such as Slack, Telegram, email, and phone. Cloud hosting is the default option, while self-hosting is available for users who want the assistant workspace and related data on their own machine. Some desktop and platform capabilities are still evolving; the Windows client was described as a development build in the consulted installation documentation.

Pricing and access

Vellum has a free Base plan that continues without a subscription fee. It includes a small hosted computer, 6 GiB of persistent storage, managed LLM credentials, and pay-as-you-go credits, but it does not include a monthly usage bundle.

Paid plans begin with Mighty at $30 per month. Super is listed at $100 per month and Ultra at $200 per month. Paid plans combine included monthly allowances with prepaid or pay-as-you-go credits, so the subscription price is not necessarily the complete cost for heavy usage. Included allowances reset each billing cycle and do not roll over, while purchased credits expire after 12 months. Custom configurations are also available, with a documented minimum configuration of $50 per month.

Vellum is primarily a personal assistant rather than a conventional enterprise collaboration suite. The supplied documentation does not describe a standard enterprise plan with shared workspaces, SSO, role permissions, or extensive administrative controls.

Privacy, permissions and limitations

Vellum documents encrypted and isolated cloud workspaces, encrypted credential handling, and a separate credential execution service. It states that customer conversational data is not used to train or fine-tune AI models, although opted-in content may be used to improve and evaluate the service. Conversation context is sent to the configured model provider for inference, and external channel providers may receive messages sent through those channels.

Self-hosting can keep workspace files, memories, conversations, credentials, and trust rules on the user's machine. This can be important for users who want more direct control over where assistant data is stored, but it may also require more technical setup and maintenance.

The broad access that makes Vellum useful also creates risks. Users should review permissions carefully, limit connected services to what is necessary, and understand approval and trust settings before allowing automated actions. Model behavior and available features can vary by provider and configuration, and usage credits can create additional costs. Vellum also lacks a native Linux desktop application and is less suitable for teams that need shared workspaces, formal administration, or enterprise permission structures.

Who should use Vellum?

Vellum is a good fit for individuals, founders, operators, researchers, and technically comfortable professionals who want one assistant to retain context and work across personal and business tools. It is particularly relevant for people whose work involves repeated research, communication, scheduling, document handling, and browser or computer tasks.

It is less suitable for an organization seeking a shared team workspace, a tightly governed enterprise assistant, or a specialized tool focused only on writing, search, analytics, image generation, or meeting transcription. Users who only want occasional question answering may find its integrations, permissions, and usage model more involved than necessary.

Bottom line

Vellum's defining feature is the combination of memory, connected tools, and controlled action-taking across multiple interfaces. It sits between a general-purpose chatbot and a personal automation system: it can converse and create content, but its larger purpose is to help manage information and complete tasks over time. The product is most compelling when continuity and cross-application action matter, provided the user is comfortable managing permissions, configuration, and usage costs.

For more products in this category, see the broader AI tools directory.