Productivity & Workflows
Capacities Review: Object-Based Notes for Work Research and Ongoing Projects
A fictional two-month orientation-sign research packet links people, meetings, sources and an open question, then prepares eight notes and two attachments for a bounded migration and exit evaluation.
Capacities is a candidate for personal work research when you need to connect a project, its sources, dated observations and the people involved without keeping independent copies in monthly folders. Define the recurring research questions, then check a small source packet before moving your archive. Basic already covers core notes, links and export; a paid plan needs a specific requirement such as saved queries or version history.
This LandOffer.ai review checked official documentation and personally inspected public pricing on October 27, 2026. Its cross-month project and migration packet are original fiction. We checked our own files and arithmetic locally; we did not create a Capacities account, import notes, connect a calendar, run AI, synchronize devices or export product data. Product retrieval quality and migration fidelity remain untested.

Choose objects for recurring relationships
Capacities organizes notes as typed objects. Built-in types include Pages, Weblinks, Images and PDFs; custom types can carry properties appropriate to a person, meeting or project. A Weblink object has more structure than an ordinary pasted URL. The object-type guide also describes separate type dashboards and linked content.
Use linked objects when you repeatedly revisit the same person, source or project on different dates. It creates maintenance work when every passing thought becomes a new type with properties nobody uses. Start with the questions you actually revisit. A folder-based notebook can remain adequate when you mostly read a linear project journal and rarely navigate between recurring entities.
The current daily-note guide describes capture followed by review: keep fleeting material on the daily page, then turn useful lines into longer notes or linked objects. Capture date, meeting date and a source's publication date can differ. Do not overwrite those distinctions merely to make a timeline look tidy.
For this review, the intended surface is a managed Windows desktop application and an Android phone for later readback. Those are evaluation requirements, not tested installations. The proposed user is one researcher maintaining personal working notes. The product's pricing FAQ says Capacities is designed for individuals and does not offer organizational plans; view-only sharing is not a shared editorial workspace with team approval controls. Pricing.
Completed model: carry one research project across two months
Fictional researcher Devon is comparing two orientation-sign layouts for the Mosaic training center. The research period runs from November 16 to December 4, 2026. Tali supplies facilities context, while the center's existing document owner retains authority over the final sign. Every person, observation, source and decision below is invented; no visitor research or personal record was collected.
Devon's question is narrow: can a person recover why the north-entrance option remains pending after a second review? Tali supplies facilities context; only an attributable decision can establish what she agreed to. The completed object specification contains eight notes and two attachments:
| Record | Type and essential content | Relationship and boundary |
|---|---|---|
| P1 Mosaic orientation | Project; period Nov. 16–Dec. 4; outcome: a comparison for review | Links the research packet; final installation approval stays outside the notebook |
| U1 Tali | Person; facilities contact in this fictional case | Linked from both reviews; a person link is not an approval |
| M1 November walk-through | Meeting; Nov. 18; north entrance has two plausible branches | Links P1, U1 and source S1; keeps the ambiguity visible |
| M2 December review | Meeting; Dec. 2; side-entrance option is acceptable within the example | Links P1 and U1; does not silently close the north-entrance question |
| S1 Layout specification v2 | Source note; synthetic Nov. 1 document; relevant passage on page 2 | Links the PDF attachment and records “page 2, entrance branch paragraph” in plain text |
| S2 Review comparison | Source note; authored comparison of the two layouts | Links the route sketch and both meeting notes; separates observation from recommendation |
| Q1 North branch | Question; Open; owner Devon; next review Dec. 4 | Links M1, M2 and S1; no confirmed answer is invented |
| D1 Review recommendation | Decision note; side option may proceed to the document owner; north remains pending | Links Q1 and S2; recommendation is not installation approval |
The attachments are a synthetic two-page specification PDF and a simple SVG route sketch. The eight note records plus two media files make ten source files in our local specimen. This is a prepared source packet, not a count observed after a Capacities import.
Devon's November daily capture can link M1 and P1 without becoming the sole home of the research. In December, M2 links the same P1, U1 and Q1 identities. Copying the full project into a December folder would create another record to reconcile. Link both reviews to P1 so a December reader can recover the November question without maintaining a second project copy.
The filled recommendation reads: “Use the side-entrance layout as the comparison's accepted option for further document review. Leave the north-entrance branch unresolved. Devon will check the page-2 requirement with Tali on December 4 and update Q1 with the attributable answer. The document owner must approve any final sign before production.” This is teaching material, not an actual instruction sent to a colleague.
Make the retrieval question more precise than a graph
The useful retrieval exercise asks three questions. First, which reviews involved Tali? The authored answer is M1 and M2. Second, why is the north option pending? Q1 refers to the ambiguous branch in M1, the still-unresolved state in M2 and the page-2 source locator in S1. Third, what may the document owner review now? D1 recommends the side option and retains the north exception.
Our executed local Python check verifies that the authored links have named targets and Q1 remains Open; it checks our source packet only. It does not measure Capacities search or prove that its backlinks render the intended passage. A real evaluation would open the same questions on desktop and phone and inspect the returned source content, not count the number of lines in a graph.
Pro's queries collect objects using rules such as types, properties, tags and backlinks. Results update as matching content changes. A saved query is valuable when the same filter must be repeated over a growing collection. Its result still depends on the data and rules you supplied.
For this packet, a proposed saved query selects Question objects linked to P1 whose status is Open. Its expected result is Q1. An unlinked question could be missing even if its title mentions Mosaic. A closed question could remain if its status was not updated. This is an expected answer key, not an executed query or vendor command syntax.
Devon can initially inspect this eight-note packet manually on Basic. Pro becomes a candidate if repeating those filters across more projects is worth paying for, or if version history becomes a requirement. The object model alone does not prove an upgrade is needed. AI analysis also adds no evidence to the unanswered facilities question unless its output is checked against an attributable source.
Completed migration review: fix ambiguity before importing

The bulk-import reference says bulk import is a desktop feature. It supports Markdown, CSV and specified media, attempts type mapping and resolves supported note links. It also warns that it is not a lossless space-migration tool. A supported format is a starting condition, not proof that the target preserved every relationship.
Our second completed asset is the preflight and acceptance record for the same Mosaic packet. The original draft had two notes both titled “Review,” one from each month. Devon's local inventory review flags that ambiguity and renames them “November walk-through” and “December review,” updating their links. No Capacities result is used to claim the ambiguity was detected automatically.
| Item checked | Completed source-packet finding | Required product readback |
|---|---|---|
| Inventory | Eight uniquely named notes and two attachment files; original copies retained | Reconcile created, skipped and failed records; do not assume ten files guarantee ten new objects |
| Types and properties | P1 Project, U1 Person, M1/M2 Meeting, S1/S2 Source, Q1 Question, D1 Decision | Inspect actual mappings, dates and Open status rather than trusting folder names alone |
| Note links | Both monthly reviews target the same P1 and U1; D1 targets Q1 and S2 | Open each link, including both renamed reviews; compare targets, not just clickable styling |
| PDF evidence | S1 names specification v2 and the page-2 paragraph in text | Open the actual attachment and locate the passage; a PDF link without its locator is incomplete evidence |
| Route sketch | S2 embeds the local SVG; filename and checksum recorded | Inspect the rendered sketch and its attachment identity on the intended device |
| Exit package | Prepared notes, media inventory and relationship answer key remain available locally | Export from an authorized sample space and repeat the three retrieval questions outside the app |
The import documentation specifically warns that a link fragment such as a PDF page selector may resolve only to the underlying file. We therefore retain the locator as text in S1 instead of relying entirely on #Page=2. The local packet's link check strips the fragment to check file existence while separately requiring the locator sentence. That is our own preflight rule, not a simulation claiming how Capacities processed the file.
The completed disposition is “eight linked notes with two review titles corrected, plus two preserved attachments; ready for a permitted sample import, with product acceptance pending.” Devon has corrected duplicate titles and preserved the source passage locator. He has not shown successful mapping, matching, media upload or round-trip restoration. Retaining originals keeps the remaining evaluation reversible without declaring a migration complete.
Treat synchronization and export as different checks
The offline guide says a device is offline-ready only when notes have downloaded. Media settings differ: browser defaults can leave some files unavailable offline. On the proposed phone, Devon must check the PDF and route sketch themselves, rather than infer readiness from a visible project title. AI features require online access.
Synchronization also differs from a recoverable copy. The sync-status guide identifies pending and failed synchronization states. A locally saved correction is not proof the second device has received it. Our review did not exercise conflict handling or synchronize any note; preserve work before any recovery action that could discard a version.
The export guide documents readable files, properties in frontmatter, collections in CSV, media folders and local links. Media must be downloaded on the exporting device to be included. Automated exports also require a selected destination and the app open at the scheduled time. These are documented conditions, not a backup task we configured.
For Mosaic, an authorized export check would open P1, trace D1 to the still-open Q1, recover M1 and M2, and inspect both attachments outside Capacities. It would compare the actual export's names and relationships with the answer key. No claim is made that every target application reconstructs the same types, queries or interactive views from Markdown.
Choose Basic or Pro after the packet fits
The public selector showed Pro at USD $11.99 monthly or $9.99 monthly equivalent billed yearly. Twelve displayed equivalents model $119.88, versus $143.88 for twelve monthly payments; the $24 difference is arithmetic before taxes and final checkout terms. Believer adds support and early-access positioning, not an organizational account for this case. Current prices.
The same page describes a light included AI budget on Pro and Believer. Plus increases it; additional top-ups require Plus. No fixed number of research questions can be inferred from that budget. Devon requires no AI workflow initially, so neither Plus nor an automatic top-up belongs in this purchasing proposal.
The completed recommendation is Basic for a bounded synthetic evaluation, with Pro considered only for a demonstrated saved-query or history requirement. Keep the existing team decision system and originals. Before subscribing or moving real research, have Devon record the actual mapping, three retrieval answers, both attachments on the phone, sync state and an independently opened export. LandOffer.ai publishes related workplace reviews; this decision remains about maintaining a usable, attributable research record.


