Liminary

• One-click capture of web pages, PDFs, videos, AI chats, and research sources
• AI Recall for retrieving relevant saved knowledge during active work
• Fact-checking that compares written claims against saved research sources
• Gap Detection for identifying relevant research missing from a draft
• Magnet meeting assistant for capturing, transcribing, summarizing, and recalling meetings
• AI Composer for creating reports, proposals, social posts, and other deliverables
• AI Agents that monitor saved knowledge and deliver scheduled briefings or alerts
• MCP connectors for connecting Liminary with external tools and knowledge sources
• Conversational AI with support for Anthropic Claude, OpenAI GPT, Google Gemini, and Amazon Nova
• Collaborative research projects with shared knowledge, sources, and team workflows

What Is Liminary?

Liminary is an AI-powered knowledge companion designed for research-intensive work. It captures information from different sources and organizes it into a searchable knowledge environment that can be recalled while users are researching, writing, meeting, or making decisions.

The platform is designed around the idea of proactive knowledge retrieval rather than requiring users to repeatedly search through their saved information. It can surface relevant sources during active work and help transform accumulated research into new outputs.

AI-Native Knowledge Storage

Liminary combines storage, retrieval, AI analysis, and research workflows in a single environment. Users can save information from websites, PDFs, videos, AI conversations, documents, and meetings and later retrieve relevant context through natural-language interactions.

The platform also maintains source relationships and contextual information so that saved research can be reused across different projects and tasks.

One-Click Research Capture

Liminary provides browser-based capture for saving research while browsing. Users can save articles, reports, web pages, PDFs, videos, and AI chat content into their library.

Users can also add personal notes and context while saving material. This allows the knowledge library to contain both the original source and the user’s own observations about it.

Knowledge Recall

Recall is one of Liminary’s core capabilities. Instead of relying exclusively on keyword searches, the system can retrieve relevant information from saved knowledge based on the context of the user’s current work.

This allows information saved previously to become available again when it becomes relevant to a new document, research project, conversation, or other task.

Proactive Recall

Liminary’s proactive recall approach is designed to surface relevant saved information without requiring users to stop working and manually search their library.

The system can recognize connections between the current task and previously saved research, helping users bring older information back into the workflow when it is potentially relevant.

Source-Grounded Answers

Liminary grounds its knowledge workflows in information that users have saved. Answers and research outputs can be traced back to the underlying sources, helping users inspect the material behind an AI-generated response.

This source-oriented approach is intended for knowledge work where users need to verify information rather than rely solely on an unsupported AI response.

Fact Checking

Liminary provides a fact-checking feature that compares claims in a document against the user’s saved research. Rather than checking only against general internet information, the Writing Sidekick can evaluate claims using the sources the user has already collected.

The system can show discrepancies between what a draft says and what the user’s saved sources contain, allowing the user to review and apply suggested corrections.

Gap Detection

Gap Detection identifies relevant information present in a user’s research that is not adequately represented in a draft. It can surface missing points, overlooked evidence, or research that may need to be considered before a document is completed.

This feature is designed to help users review the completeness of research-based writing before publication or delivery.

Writing Sidekick

The Writing Sidekick works alongside documents and web-based writing environments. It can identify claims that may need verification and help users compare their writing with saved research.

The tool currently supports workflows involving Notion, Google Docs, and documents on the web through the Liminary browser extension.

AI Composer

Liminary’s Composer helps turn saved knowledge into finished deliverables. Users can ask the system to create outputs such as social posts, proposals, brand guides, research reports, and other documents.

The workflow can ask users a series of questions before generating the artifact, allowing the output to be shaped around the intended audience, purpose, and available research.

Google Docs Integration

Created artifacts can be synchronized with Google Docs. Liminary describes a two-way workflow in which users can build and refine content in Liminary while keeping the document synchronized with Google Docs.

Meeting Assistant

Magnet is Liminary’s meeting assistant. It captures meetings and makes the resulting information available for later recall.

The current product includes meeting summaries, full transcripts, and notes that can be saved into the user’s knowledge library for future research and reference.

Live Meeting Assistance

The Live Meeting Assistant can recall relevant knowledge while a meeting is taking place. This gives users access to previously saved information during conversations rather than only after a meeting has ended.

The resulting meeting information can also become part of the user’s long-term knowledge library.

AI Agents

Liminary provides AI Agents that can operate on schedules rather than waiting for individual prompts. Users can configure an agent to run daily, on weekdays, or on selected days.

Agents can monitor saved knowledge and surface changes, summaries, briefings, or alerts based on the user’s configured workflow.

Scheduled Research

Agents can be configured to monitor specific projects or knowledge areas. This allows Liminary to track new information and provide scheduled updates instead of requiring users to repeatedly perform the same research task manually.

MCP Connectors

Liminary supports Model Context Protocol connectors that allow its knowledge environment to interact with external applications and services. Connectors can bring information from tools such as Slack, Notion, Gamma, Granola, and Zoom into AI-assisted workflows.

The platform also provides an MCP server that can allow compatible AI assistants to search and work with a user’s Liminary library with appropriate permissions.

Claude Integration

Through its MCP server, Liminary can connect with Claude so that Claude can search and recall saved documents, notes, and meetings. Depending on permissions, Claude can also save or edit content in the Liminary library.

Users can configure read and write permissions for individual Liminary tools, providing control over which actions an external assistant can perform.

AI Model Selection

Liminary provides model selection for chat on its Pro plan and above. Current supported model families include Anthropic Claude, OpenAI GPT, Google Gemini, and Amazon Nova.

The company states that it evaluates models using internal benchmarks and can add or replace models as performance changes.

Research Synthesis

Liminary is designed not only to store research but also to connect information from different sources. It can help identify relationships between documents, notes, meetings, and other saved material.

This makes the platform suitable for research workflows where insights depend on comparing information across multiple sources rather than reading a single document in isolation.

Watchlists

Users can create watchlists for clients, projects, or research topics. Liminary can monitor the relevant knowledge area and surface new information or changes through its agent-based workflows.

Collaborative Projects

Liminary supports collaboration for teams working on shared research. Pro users can collaborate with up to 10 people at a time per project, while the Team plan supports sharing with up to 50 people per project.

The Team plan also includes a shared team knowledge base for maintaining research and organizational context.

Team Knowledge Base

The Team plan provides a shared knowledge base where research and other saved information can be used across team projects. This allows teams to maintain common context instead of keeping research isolated in individual accounts.

Workflow Integrations

Team users receive workflow integrations that currently include services such as Google Drive and Granola. These integrations extend Liminary’s knowledge environment into other tools used for documents, meetings, and collaborative work.

Supported Knowledge Sources

Liminary supports a broad range of knowledge sources, including LLM chats, web pages, PDFs, videos, documents, spreadsheets, and meeting information. The platform is designed to bring these sources together so they can be recalled and analyzed through the same knowledge environment.

Browser Extensions

Liminary provides browser extensions that allow users to capture and interact with research while browsing. The current product is available for Chrome and Firefox.

The extension also supports the Writing Sidekick workflow, allowing Liminary to operate alongside supported online documents.

Privacy and AI Data Handling

Liminary states that user content is encrypted and siloed to the user’s workspace. The company also states that user content is not used to train generative AI models.

Liminary uses AI models to provide features such as summarization, question answering, recall, synthesis, and analysis of saved content. The platform states that its models run in its own cloud environments or through APIs under enterprise agreements with model providers.

Free Plan

The Explorer plan is available for free. It includes up to 200 saved documents and web content, 10 local uploads per day, three Composer uses, basic recall, 30 days of chat history, and the ability to save up to 10 meetings with Magnet.

Pro Plan

The Pro plan costs $29 per month. It increases the storage allowance to 10,000 saved sources and provides up to 100 local uploads per day.

Pro also adds advanced recall with fact-checking and gap detection, unlimited saved Magnet meetings, model selection, 24 months of chat history, MCP Connectors and Agents, and collaboration with up to 10 people per project.

Team Plan

The Team plan costs $50 per user per month. It includes everything in Pro along with unlimited sources and uploads, a shared team knowledge base, collaboration with up to 50 people per project, workflow integrations, advanced AI models, custom onboarding, unlimited chat history, and a dedicated support channel.

Who Is Liminary For?

Liminary is designed for people whose work involves collecting, analyzing, connecting, and communicating information. Its use cases can include research, consulting, strategy, analysis, professional writing, knowledge management, and meeting-heavy workflows.

For Researchers

Researchers can use Liminary to collect information from multiple sources, maintain notes alongside original material, retrieve previous research, and synthesize information across documents and other media.

For Consultants

Consultants can maintain project-specific research collections and use saved information when preparing recommendations, reports, presentations, or client communications. Team functionality can also support shared project knowledge.

For Writers

Writers can use the Writing Sidekick to compare claims against their saved research and identify information that has not yet been incorporated into a draft. Composer can also help transform research into structured deliverables.

For Teams

Teams can use shared projects and team knowledge bases to maintain common research. External integrations and MCP connectors can extend that knowledge into other tools used within the team’s workflow.

For Meeting-Heavy Work

Magnet allows meeting information to become part of the same knowledge environment as documents, web research, and notes. This creates a searchable source of meeting information that can be recalled during later work.

Conclusion

Liminary combines AI-powered knowledge storage, proactive recall, research capture, fact-checking, gap detection, meeting assistance, document creation, AI agents, and external connectors. Its free Explorer plan provides a limited version of the workflow, while Pro and Team expand storage, collaboration, automation, integrations, and AI capabilities.

FAQ: What is Liminary used for?

Liminary is used to capture, organize, recall, analyze, and synthesize research and other knowledge from sources such as web pages, PDFs, videos, AI chats, documents, and meetings.

FAQ: Is Liminary free?

Yes. Liminary has a free Explorer plan with limits on saved sources, uploads, Composer usage, chat history, and meetings.

FAQ: How much does Liminary cost?

The Pro plan costs $29 per month, while Team costs $50 per user per month. The Explorer plan is free.

FAQ: What is Liminary Recall?

Recall retrieves relevant information from the user’s saved knowledge and can surface it in the context of ongoing work instead of requiring a traditional manual search.

FAQ: Can Liminary fact-check documents?

Yes. Its Writing Sidekick can compare claims in a document against saved research and identify discrepancies for review.

FAQ: What is Gap Detection in Liminary?

Gap Detection identifies relevant information in saved research that may be missing from a draft, helping users review whether important points or evidence have been overlooked.

FAQ: Does Liminary record meetings?

Liminary’s Magnet meeting assistant can capture meetings and save summaries, transcripts, and notes for later recall.

FAQ: Does Liminary support AI agents?

Yes. Liminary Agents can run on schedules and monitor saved knowledge to deliver briefings, summaries, alerts, or other configured outputs.

FAQ: Does Liminary support MCP?

Yes. Liminary provides MCP connectors and an MCP server that can connect its knowledge library with compatible AI assistants and external services.

FAQ: Which AI models does Liminary support?

Liminary currently supports model families from Anthropic Claude, OpenAI GPT, Google Gemini, and Amazon Nova for chat, with availability depending on the user’s plan.

FAQ: Can teams use Liminary?

Yes. Pro supports collaboration with up to 10 people per project, while Team supports sharing with up to 50 people per project and adds a shared team knowledge base.

Share Your Experience: