Liner combines web research, an academic search experience and drafting in one workspace. Its most important distinction for serious users is the chain from question to individual source, not the speed of the generated answer.
What this product actually is
The current official Liner site presents Search, Scholar and Write as complementary experiences. The consumer product was previously associated with highlighting; this article concerns the research workspace as of October 2026, not every historical feature or specific underlying model release.
Who it can help and when
Liner can be useful for an undergraduate building an initial reading list, a policy researcher collecting reports or an analyst turning sources into a first outline. The research task is finding inspectable sources with traceable metadata, rather than accepting an AI-generated references section at face value.
A realistic first workflow
Write an explicit research question and ask for sources from a bounded year range. Open each cited work, check DOI, authors and actual methodology, then create a table distinguishing primary studies, reviews and commentary. Make the final narrative independently traceable even if the assistant changes the wording in subsequent drafts.
Limitations and decisions the user must make
An apparently scholarly answer may misclassify a preprint as peer reviewed, mistake a paper abstract for its findings or omit contrary research. Results depend on indexing and rights. Separate Search and Scholar availability from writing functions; do not assume all models and features are on every subscription.
Privacy, permissions and data handling
Liner’s July 21, 2026 privacy update and its March 30 regional policy treat jurisdictions differently and describe content processing and training choices. A user should examine their applicable regional policy and settings before uploading unpublished manuscripts or personal research notes.
Portability, export and avoiding lock-in
Store DOI lists, BibTeX records and annotated tables outside Liner. A generated bibliography does not guarantee archival source availability. Any full-project export claim is left unverified pending a specific dated export guide.
Alternatives and adjacent tools
Semantic Scholar supplies scholarly discovery, SciSpace provides paper reading, and Scite focuses on citation context. These are different evidence operations, not a ranking of AI intelligence.
Status, availability, and account-specific caveats as of October 25, 2026
This is a first-party documentation profile, not an independently tested or objectively rated product. Account access, supported languages, accessibility support, model selection, data residency, enterprise entitlements and live workflows can differ by country, organization and client version. Where a precise entitlement cannot be verified against a dated first-party source for a specific region and tier, it remains unknown.
Pricing, testing and editorial boundary
No current numeric subscription price, usage quota, API token cost or model benchmark score is asserted in this guide. Those require separate dated primary-source snapshots managed by a dated pricing reference. This profile is not an endorsement, independent product test, legal/security audit or permission to copy screenshots, branding or proprietary documents. Deep model architecture, code agents and imaging features are covered by their respective specialist editorial teams.
Sources & Further Reading
Official first-party documents checked for this editorial build on October 25, 2026; date-bound features must be reevaluated after the cutoff. Vendor statements have not been independently performance-tested.