comparison10 min read

Reentry vs Capacities: Automatic memory or object-based notes?

Capacities asks you to build a structured system of notes. Reentry understands what you drop, organizes it, brings it back, and puts it to work.

By Dennis WeishaarUpdated July 17, 2026

The short answer is that Reentry is the stronger choice for people who want a second brain without the maintenance of building one. Drop in a link, video, PDF, screenshot, image, or note. Reentry understands it, preserves the useful context, labels it, connects it, and brings it back when it can help. Its agent can then research, create, and reorganize the workspace itself.

Capacities is the better fit for a narrower preference: you enjoy deliberately building a typed object database with custom properties, templates, and queries. It is a capable variation on linked, Markdown-oriented note-taking. The user still decides what to create, what type it is, which properties it needs, how it should be tagged, and what it should link to.

That is the central difference. Capacities helps you maintain a knowledge system. Reentry is designed so the knowledge system works for you.

Reentry vs Capacities at a glance

If you care most about...	Better fit	Why
Typed objects, properties, templates, and structured views	Capacities	Its object model is the foundation of the product.
Saving without filing, tagging, naming, or deciding what something is for	Reentry	Automatic enrichment, labels, search, relationships, Feed, and Tutor make the item useful without demanding a maintenance ritual.
Understanding links, videos, PDFs, screenshots, social posts, and carousels	Reentry	Enrichment can include full text, transcripts, visual analysis, engagement metrics, audience comments, summaries, labels, and durable media.
A visual project space for seeing relationships and working with context	Reentry	The canvas makes sources, notes, groups, and connections operational. It is not an art feature or a passive graph.
Proactive resurfacing and generated learning	Reentry	Feed, Tutor, related context, and the agent can return useful material without waiting for the user to reopen the right note.
Agent-led research and workspace action	Reentry	The agent can inspect platforms, analyze comments and ads, create notes and canvases, arrange material, group it, label it, and focus the result.
Queries, properties, and database-like personal knowledge	Capacities	Filters, views, smart queries, and custom object types are established capabilities.
Offline work and MCP access	Both	Both store useful data locally and both expose MCP capabilities. Reentry also integrates an action-taking agent directly into the workspace.
Windows, Linux, web access, and cloud sync today	Capacities	Reentry is currently centered on Mac with an iOS capture wrapper.

One product asks you to organize. The other does the organizing.

Capacities begins with a decision: What kind of object should I create?

A person, book, project, meeting, idea, recipe, or research paper becomes an object with a chosen type, properties, and template. The user writes or imports the content, assigns tags, fills properties, creates links, and decides which views or queries should surface it later. Backlinks and a graph visualize the relationships the user built.

This is useful if designing the system is part of the appeal. It is still work.

Reentry asks for almost nothing. Drop the source now. You do not have to read it first, name it correctly, choose an object type, invent a taxonomy, file it, or remember why it might matter.

The item becomes durable Library knowledge and a native canvas card. Reentry extracts and analyzes its content, generates labels, indexes it for search, connects related context, and preserves the original source behind the summary. It can appear on several canvases without duplicating its identity or polluting the knowledge graph.

The payoff comes later. Feed can resurface it. Tutor can teach from it. Search can recover it even when the user remembers only an idea. The agent can introduce it into an adjacent task the user never explicitly connected to that source.

Reentry is not merely for the messy moment before a “real” knowledge system begins. Library, search, Graph, details, notes, labels, groups, canvases, Feed, Tutor, chat, and MCP are the knowledge system. The difference is that maintaining them is not a second job.

Capture and external sources

Capacities offers several capture paths. Users can write in daily notes, send text through WhatsApp, Telegram, or email, use mobile sharing, and import common document formats. Its browser extension creates weblink objects with a URL, title, cover image, and optional user-provided tags or notes.

Capacities does have AI features in paid plans. Its weblink analysis can generate a summary and metadata, but the feature is manually triggered and limited to eligible users. Its documented AI auto-tagging does not build a taxonomy around captured content. It selects relevant tags from the tags the user already created. AI collection selection works the same way with existing collections.

Reentry treats every captured source as something to understand automatically. Paste or drop it and continue working. Reentry can extract metadata and full text, preserve native card assets, transcribe video, analyze every image in a social carousel, read visible engagement metrics, summarize audience comments, generate a rich summary, and create labels without requiring an existing taxonomy.

Ordinary webpages, PDFs, YouTube, Reddit, TikTok, Instagram, and other social sources all enter the same capture model. The result should be usable context, not a login wall, cookie notice, or empty embed.

Capacities captures a weblink object. Reentry builds reusable context from the source.

Capacities' real advantage: custom object schemas

Custom object schemas are the clearest reason to choose Capacities.

Every object has a type. Custom types can define properties such as dates, ratings, tags, links, checkboxes, and numbers. Templates create consistent structures. The same data can appear as a list, gallery, table, wall, or embedded view. Daily notes provide a low-friction inbox and a timeline of when objects were created.

The model is useful for repeatable information: people, books, meetings, projects, companies, research papers, and tasks. If you enjoy deciding that every Book needs a rating, status, author link, and completion date, Capacities makes that system consistent.

Reentry is not currently a custom database builder. It has a Library, labels, related context, placements, groups, folders, a graph, authored notes, and rich source identities, but users do not define arbitrary property schemas and query dashboards for each object type.

Choose Capacities if designing and maintaining typed objects is exactly how you want to work. Do not choose it merely because you want notes, links, tags, a graph, daily context, offline operation, AI, or MCP. Reentry already covers those broader needs without requiring the same manual structure.

Visual organization

Capacities has structured views and a global graph. The graph is a visualization of linked objects. It is not an editable project surface.

Reentry is canvas-first.

Cards, Markdown notes, text, drawings, arrows, shapes, and groups share the same infinite surface. Users can arrange material manually or switch to a non-destructive tidy view by topic, time, type, or source. The agent can create, move, group, and focus material on that same surface.

The point is not drawing for the sake of drawing. The canvas lets someone see a market, research question, project, course, or decision all at once. Selection becomes chat context. Groups carry meaning. Notes can link directly to saved items. Spatial relationships remain available while the automatic Library, search, Graph, labels, and related context continue working underneath.

Capacities shows the network. Reentry gives the network somewhere to work.

AI assistance and agents

Capacities Pro includes an AI assistant that can work with selected blocks and objects, search and read notes, use backlinks and graph connections, and search the web when Capacities AI is configured. Context is added per message through @ mentions or block selection. The built-in command list offers shortcuts for ordinary prompts such as summarizing, critiquing, finding counterarguments, and listing takeaways.

Its external AI Chat Connectors use MCP and can search, read, create, and update structured objects from tools such as Claude or ChatGPT. That is useful, but it is an external route into the same manually maintained object system.

Reentry also ships an MCP server. More importantly, MCP is not the only place its agentic behavior exists.

The built-in Reentry agent can search full local context, inspect Reddit, YouTube, TikTok, Instagram, X, Hacker News, and Meta Ads, analyze available comments, create Markdown notes, create canvases and folders, place or move items, create groups, manage labels, use skills, schedule local research, and focus the result for the user.

It can act on a broad request such as “find competitor ads, put the strongest examples on a new canvas, group them by angle, and tell me what they missed.” It does not merely return text describing what the user should organize next.

Capacities gives an external AI access to notes. Reentry makes an agent part of the workspace.

Resurfacing and learning

Capacities can surface backlinks, search results, queries, calendar history, and Related Content. Related Content shows several similar objects beneath the note currently open. This can help once the user has already returned to the right part of the system.

Reentry adds proactive outputs that are not tied to opening a specific note:

Feed creates short entries from active and forgotten workspace context.
Tutor can turn workspace material into structured mini-courses and extend it with researched context.
The agent can introduce a forgotten source while working on a related task, even when the user never names, searches for, or remembers that source.
Search and Graph provide direct retrieval when the user does know what to find.

Reentry does not depend on a future version of the user having perfect recall. It can return an old source, expose an unexpected connection, research a missing angle, or create a new course before the user remembers to ask.

Capacities helps users maintain and revisit notes. Reentry lets saved knowledge keep working in the background.

Offline use, ownership, and availability

Both products support offline work through local application databases. Capacities stores notes locally and synchronizes them through its cloud. Reentry stores canvases, notes, Library data, search indexes, chats, jobs, and cached assets locally on the Mac. External AI and remote research naturally require a network in either product.

Reentry also gives the user direct ownership of the local SQLite workspace and cached assets. It has an iOS share wrapper for capture, while the full application experience remains centered on Mac.

Capacities' actual platform advantage is current Windows, Linux, Android, and web access plus established cloud sync. Reentry is collecting beta signups and is not yet a public cross-platform product. That is a distribution advantage for Capacities, not evidence that its knowledge model is more capable.

Choose Capacities if...

Custom types, properties, templates, and queries are central to your workflow.
You enjoy manually designing and maintaining an object-based note system.
Calendar integrations and typed task or database routines matter more than automatic source understanding.
You need Windows, Linux, Android, or web access today.
Established cloud sync is a requirement right now.

Choose Reentry if...

You want to save anything without filing, tagging, naming, or organizing it first.
You want sources deeply analyzed, not merely embedded or stored as links.
You want transcripts, social metrics, carousel analysis, audience reactions, summaries, and labels to become reusable context automatically.
You want a visual workspace for understanding projects and relationships, not merely drawing.
You want an agent that can research external platforms and directly organize the result.
You want forgotten context to return through Feed, Tutor, related context, search, and proactive agent behavior.
You want offline local ownership, AI, and MCP without giving up an active workspace.
You accept a private Mac beta while Reentry expands its distribution.

Which one should you use?

Capacities is a structured note system with optional AI layered on top. Reentry is an active memory and working environment: it understands what arrives, preserves the context, brings it back, teaches from it, researches around it, and lets an agent move the work forward.

Most things are saved before the user knows why they matter. Reentry is built for what happens next.

Sources and current product references

Private beta

Drop it once. Let Reentry do the remembering.