Marqly Review 2027: Features, Pricing, Pros & Cons

Marqly is an AI bookmark manager whose central bet is retrieval rather than filing: on Pro it auto-tags and summarises everything you save, and semantic search finds a save by describing what you remember instead of matching the words in a title. It runs on the web, in Chrome, Edge, Firefox and Safari extensions, and on iOS and Android, all against one library. The honest summary is that it is strong at finding things and weak at being free — the free plan stores 100 bookmarks and contains no AI at all, so the feature this product exists for is the feature you have to pay for.

This is a first-party review. Marqly publishes it, which means you should read the cons section first and treat the pros as claims from an interested party. What we can offer in exchange is specificity: every limit below is stated as a number or a platform name rather than a vibe, and the changelog section lists the ways the product got worse for some users over the past year, which is not something a marketing page usually volunteers. For independent framing, see how we test and the 2027 bookmark manager index.

What Marqly is

Marqly is a save-anything, find-it-later tool. The extension puts a one-click save button in the toolbar, a save-all-tabs button for clearing a research binge, and a side panel (Chrome and Edge) with Bookmarks, Clipboard, Highlights and AI Chats workspaces beside whatever page you are on. Firefox uses the extension popup; Safari does not have the Chrome-style side panel.

The organising idea is that filing is the part people abandon. On Pro, saves are tagged and summarised automatically, so the library sorts itself and a backlog becomes skimmable. Underneath that sits the retrieval layer: semantic search matches meaning across titles, page content, highlights and YouTube transcripts, so “that video about sourdough starters” works even when the title says nothing of the kind.

Platforms, stated plainly because this is where older descriptions of Marqly went wrong: web app, Chrome extension, Edge extension, Firefox extension, Safari extension, iOS app and Android app. There is no public API and no self-hosted deployment.

Key features

Capture. One-click save; save with tags; save-all-tabs; X/Twitter bookmark capture from the extension; a bookmark button in YouTube’s own action row that saves the video with its transcript attached.

YouTube AI card. On every watch page: a Summary tab with a streaming TL;DR, key sections and books mentioned; a Transcript tab that is playback-synced with one-click copy; and a Chat tab that answers questions grounded in that video’s transcript on Pro. Free accounts get a YouTube-summary allowance of three per month.

Highlighting. Select text on any website and highlight in six colours — free on every plan, with no gate. Notes on highlights are capped at 10 on free and unlimited on Pro. Highlights persist on the page when you return and sync to your library.

Boards and sharing. Boards group links and highlights, and a board can be published as a public web page that opens with no signup — free on every plan, view-only for visitors. This is the single most useful feature for teachers handing a resource list to a class, and it is not paywalled.

Ask (Pro). A multi-turn assistant in a right-hand panel of the web app (⌘J or Ctrl+J). It answers from your own library — bookmarks, highlights, notes and transcripts — streaming, with numbered citations that open the source. It also surfaces proactive findings on the empty state: dead links, exact-URL duplicates, forgotten topics and overlapping boards. It can change the library (tags, boards, renames, trash, merges), with every change presented as a proposal card you approve before it applies and undoable for 30 days. Three write modes control autonomy: Off, Ask me first (the default), and Small changes automatically, where deletions still ask. It remembers a few small preferences that are visible and clearable — it does not remember your whole library — and it runs on monthly credits whose limits we do not publish.

AI Organizer (Pro). Bulk organisation of an existing library, which is the answer to “can it clean up a pile I already have.”

ChatVault (Pro). Saves conversations from ChatGPT, Claude and Gemini into a dedicated AI Chats workspace. Available in Chrome, Edge and Firefox; not Safari. ChatVault, AI Chats and AI conversation capture are one feature with three names, not three features.

Also on Pro: clipboard cloud sync, broken-link checking, smart sorting, unlimited notes, and priority email support. Connected Apps exposes the same library tools to Claude, ChatGPT, Cursor and VS Code over MCP, read-only by default with a per-app toggle to allow changes and a revoke option.

Read mode and save-as-PDF are free on every plan. Save-as-PDF captures a page screen-faithfully including lazy-loaded images, processed locally in Chrome and Edge; Firefox and Safari fall back to the browser’s print-to-PDF.

Free versus Pro

	Free	Pro
Bookmarks	Up to 100	Unlimited
Search	Whole-library keyword	Keyword + semantic
AI auto-tagging	❌	✅
AI summaries	❌	✅
Ask assistant / AI Organizer	❌	✅
YouTube	3 summaries per month	Summaries + chat
ChatVault	❌	✅
Highlighting	✅ 6 colours, 10 notes	✅ 6 colours, unlimited notes
Boards + public sharing	✅	✅
Read mode, save-as-PDF	✅	✅
Offline reading	❌	✅ Web + iOS, per-device
CSV export	100 most recent	Whole library, or new-since-last-export
Import file size	10 MB	30 MB
Import auto-tagging	❌	✅

The free plan is a real manual bookmark manager, not a crippleware demo — saving, boards, public sharing, highlighting, read mode and keyword search across everything you have saved all work, with unlimited devices and no card. What it does not contain is any AI. If semantic search is why you are reading this, it is not on the free tier.

Pricing

• Free — $0, no card. Up to 100 bookmarks, whole-library keyword search, boards, public board sharing, six-colour highlighting (10 notes), read mode, unlimited devices.
• Pro — $72 per year (about $6 per month billed annually) or $9 per month billed monthly. Unlimited bookmarks and every AI feature.
• Standing first-year offer — $49 for year one, then $72 per year. That is 31% off the first year against the $72 annual price.
• Verified students — $48 for the first year, confirmed with a university email at checkout. That is 33% off the first year. There is no teacher or educator discount.
• Teams — $9 per seat per month or $72 per seat per year (about $6 per seat per month), with a three-seat minimum, so from $27 per month or $216 per year. One bill paid by the Owner, in USD. Every seat includes all Pro features plus a shared workspace with shared boards and bookmarks; each purchased seat also adds twice an individual Pro’s monthly AI allowance to one shared pool that resets monthly with no rollover. Roles are Owner, Admin and Member. Web only, no SSO or SCIM, no per-board permissions, no invoices or bank transfer.
• No trial of any length. Pro is billed the moment you upgrade. Cancel any time and the account continues on the free plan with every bookmark intact; a cancelled team works to the end of the paid period, then turns read-only with nothing deleted.

For comparison at the same price points: Raindrop.io Pro is $28 per year with unlimited free bookmarks but no semantic search at any price, Readwise Reader is $119.88 per year with no permanent free tier, and mymind’s AI tiers start at $79 per year. Marqly sits in the middle on price and is the only one of those whose paid tier changes how retrieval works.

Import and export — what actually happens

Import accepts browser bookmark HTML from Chrome, Edge, Firefox and Safari; Raindrop HTML (its JSON export is not accepted); the list.csv inside a Pocket export archive — not Pocket’s HTML preview file, which is a <ul><li> layout rather than the Netscape format and does not parse; generic CSV; and .txt, which is not exposed in the UI. Files cap at 10 MB on free and 30 MB on Pro, with 10,000 bookmarks per file.

What survives an import: title, URL, description or note, tags, and folder-to-board structure flattened to at most two board levels. What does not: original save dates (imported items take the import date), highlights, article text, and read or archive status. No de-duplication runs during import, so duplicates come through — the Duplicate Bookmark Finder is the cleanup step. Auto-tagging an import is a Pro feature; on free your existing tags arrive and nothing new is added.

Export is a real CSV button in the sidebar or profile menu. Free exports the 100 most recent bookmarks; Pro exports the whole library and can limit the file to items added since your last export. The CSV carries URL, title, description and tags — nothing else. Highlights, notes and boards are not in it; for a complete copy of your account including those, email support@marqly.com. We have published the measured behaviour of import formats, including which real files parse and which do not, in the import-fidelity study.

Pros

• Semantic search across titles, page content, highlights and YouTube transcripts — the retrieval case most of the category treats as secondary.
• Zero filing cost on Pro: auto-tagging and summaries mean the library organises itself.
• Public boards are free and open with no account, which solves the teacher-and-class problem without provisioning anyone.
• YouTube is a first-class citizen: summary, playback-synced transcript and transcript-grounded chat, with the transcript kept on the save.
• Ask can act on the library and be undone — proposal cards, three write modes, a 30-day undo window.
• Six-colour highlighting and read mode are free, with notes persisting on the page.
• Broad platform coverage: web, four browser extensions, iOS and Android on one library.
• Cheap relative to the premium end: $72 per year against Readwise Reader’s $119.88.

Cons (the honest part)

• The free plan has no AI at all and caps at 100 bookmarks. The defining feature is paywalled, and Raindrop.io gives unlimited free bookmarks.
• No public API and no self-hosting. Your library lives in Marqly’s cloud, full stop.
• The CSV export is partial. URL, title, description and tags only; highlights, notes and boards need a support request. That is a real portability limitation and you should know it before committing years of saves.
• Offline reading is narrow. Pro only, web and iOS only, per-device with no sync, no video caching, images over 2 MB and pages over 5 MB may be skipped, and nothing on Android or in the extensions.
• Import loses things. Original save dates are discarded, folders flatten to two levels, and auto-tagging the result costs Pro.
• Duplicate detection is exact-URL matching, not AI. Same page at two URLs with different tracking parameters will not be caught by normalisation alone.
• Ask’s credit limits are not published, so you cannot plan heavy assistant use against a known budget.
• No educator discount, no SSO or SCIM, no per-board permissions, Teams is web only, and there is no invoice or bank-transfer billing — which rules it out for some schools and districts outright.
• It is not a note-taking app. There is no original-writing surface, so pairing it with Obsidian or Notion is normal rather than a workaround.
• Younger and smaller than the incumbents, so the community and third-party ecosystem are thinner than Raindrop’s or Readwise’s.

What changed over the past year

Because a review is only useful if it is current, here is what actually moved — including the changes that were bad news for some users:

• The free bookmark cap went from 2,000 to 100 (September 2026). This is a substantial reduction and the single most important change on this list. Older articles and third-party reviews still quote 2,000; they are wrong.
• The free trial was retired (September 2026). There is no trial of any length now; the free plan is the evaluation path.
• Offline reading shipped (September 2026) as a Pro feature of the web and iOS apps. Before that, offline reading did not exist in the product at all — so both the old denial and any current claim of a synced offline library are wrong.
• A self-serve CSV export button shipped, replacing an email-support-only path. A complete account copy still requires support.
• Marqly Teams went public (September 2026) with shared boards and bookmarks, three roles and pooled AI.
• The first-year offer moved from $39 to $49 (September 2026), now applied with the code STANDING49.
• Ask became a real multi-turn assistant that can change the library through undoable proposals, replacing an earlier single-shot Q&A.

Who Marqly is for

People with a large, messy backlog they cannot search. Former Pocket users who want the retrieval Pocket never had — see the Pocket alternatives guide. Teachers who need students to open a resource list with no account, which is covered in detail in the teacher roundup. Students who will pay $48 for a first year of semantic search at citation time, per the student guide. Anyone building a second brain from what they read without maintaining a folder taxonomy.

Who it’s not for

If you need unlimited free bookmarks, Raindrop.io is the better answer and its free tier has no cap. If you highlight heavily and want spaced-repetition review, Readwise Reader is built for that and Marqly is not. If owning your data on your own server is non-negotiable, Linkwarden or Karakeep are open source and Marqly never will be. If your archive is documents and scans rather than links, Evernote’s OCR search beats anything here. If you want a place to write original notes, use Obsidian or Notion and let Marqly be the input layer. And if you have about forty bookmarks and never lose them, your browser’s own bookmarks are the right tool and cost nothing — see the free-tier comparison for why.

The bottom line

Marqly is a retrieval product wearing a bookmark manager’s clothes. Judged as a filing cabinet it is unremarkable and its free tier is stingy next to Raindrop’s. Judged on the question it actually answers — can you find something you saved eight months ago by describing it — it is the strongest option we know at this price, and the YouTube transcript handling and the undoable Ask assistant are genuinely differentiated rather than table stakes.

The trade is that none of it is free, the export is partial, and the data lives on our servers. If those three are acceptable, start on the free plan, import what you already have, and run your own five lost saves through it before paying anything — the protocol is in the retrieval benchmark. If they are not, the compare directory has head-to-head tables against every tool mentioned above, and we would rather you picked the right tool than ours.

---

Features, plan boundaries and prices verified against the live product and our dated competitor dataset in October 2026. This is a first-party review written by Marqly; the cons section is not abbreviated. We publish no numeric rating for our own product — a self-awarded score is not evidence, and Google does not treat it as one either.