Liner

The reason general assistants are distrusted in research is specific and well earned: they invent citations. A reference that looks correct, formatted properly, attributed to real authors, and describing a paper that does not exist.

Liner's design treats the citation as the output rather than the decoration. Every answer links to what it came from, and checking is one click.

It then splits into three agents rather than one, because finding a fact, reviewing a literature and drafting a report are genuinely different jobs.

Liner: Three Agents For Professional Work

• Three separate agents — everyday search, academic research, and professional drafting — rather than one general assistant.
• Every answer cited, with one-click access to each source and follow-up questions to go deeper.
• Liner Scholar finds and compares papers, surfaces citations and builds literature reviews.
• Mind maps for processing complex topics, alongside the linear answer.
• An API and MCP server, plus published case studies from named universities and researchers.

Why Do Citations Matter More Than Answers?

Liner's search agent is described as giving cited answers, with every citation checkable in one click and follow-up questions to deepen understanding.

That ordering — citation first, answer second — is the whole argument, and the reason is about what a professional actually needs.

An answer you cannot trace is not usable in any context where you are accountable for it. A consultant putting a figure in a client report, an analyst supporting a recommendation, a journalist publishing a claim — each of them needs the source more than the summary, because the source is what survives being challenged.

Its own displayed testimonials say exactly this. A marketing consultant describes using it to fact-check insights for client reports; a content researcher describes verifying sources before publishing. Both are checking rather than asking.

The deeper point is that a cited answer changes what the tool is for. Uncited, it is something that tells you things. Cited, it is something that finds things — and finding is the part that was actually slow.

An answer you cannot trace is unusable anywhere you are accountable for it. The source is what survives being challenged.

Liner Scholar And Finding Real Papers

The academic agent is described as finding and comparing papers, strengthening a paper with the right citations, and producing literature reviews in minutes. Among the published case studies is a professor using it to find papers without fake citations.

That phrase names the problem that has kept general assistants out of academic work entirely.

Fabricated references are the single most damaging failure a research tool can have, because they are indistinguishable from real ones at a glance. Correct author names, plausible journal, plausible year, and no such paper — which a supervisor discovers rather than the student.

A tool that retrieves from an actual index cannot invent a reference, because everything it returns exists by construction. That is an architectural guarantee rather than a quality improvement.

What it cannot do is judge the paper. Finding twenty relevant papers in minutes is genuinely valuable and it moves the bottleneck rather than removing it — reading them, assessing methodology and deciding which ones matter remains the work.

Search, Research And Write As One Window

The writing agent is described as searching, writing and editing in one window — drafting from a topic, shaping ideas into emails, newsletters and reports, and editing and verifying without breaking flow.

Keeping verification inside the drafting window is the useful part, and it addresses a specific failure in how people write with AI.

The normal pattern is fragmented: generate a draft in one tool, notice a claim you are unsure about, open a browser, search, lose your place, come back and have forgotten what you were writing. The verification step is where the flow breaks and, more often, where it gets skipped.

Collapsing the two means checking a claim costs seconds instead of a context switch, which is the difference between verifying and intending to verify.

The trade is scope. A dedicated writing tool will out-feature this on style and structure, and a dedicated editor will out-feature it on clarity — the reason a tool like Hemingway Editor still has a place beside it. What Liner offers instead is a draft whose claims you can stand behind.

• Read the citation, not just the answer — that is the whole point.
• Use Scholar for literature and the search agent for facts.
• Treat found papers as a reading list, not a conclusion.
• Check the mind map when a topic will not sit in your head.
• Verify claims while drafting rather than intending to afterwards.

Mind Maps And Structuring A Topic

Among the search features is processing complex topics with mind maps, offered alongside the conventional answer.

Visual structure is an odd thing for a search product to build, and it is more defensible than it first appears.

A linear answer imposes an order that may not match the subject. When you are trying to understand something with many interacting parts — a regulatory landscape, a technical architecture, a field with competing schools — prose forces a sequence and hides the relationships.

A map shows what connects to what, which is usually the thing you were actually trying to work out. It also makes gaps visible in a way paragraphs do not; an empty branch is obvious, a missing sentence is not.

Its displayed testimonial for this feature comes from a product manager breaking down complex topics while planning features, which is the right use. It is a comprehension aid rather than a deliverable, and it earns its place by getting you oriented faster. For general questions Komo is the lighter tool, and Daloopa applies the same source-linking discipline inside a single domain.

Who The Case Studies Are Actually About

Liner publishes case study material naming a university pharmacy programme using Liner Scholar in research education, a public policy researcher short of time, a professor avoiding fabricated citations, and a piece about a major developer community positioning itself as a knowledge layer for AI agents.

Read as a set, those describe a specific customer that most AI search products do not target.

Every one is someone whose output is checked by other people. A pharmacy educator, a policy researcher, a professor — all working in contexts where a wrong citation has consequences beyond embarrassment.

That explains the product decisions. Citation-first output, an academic agent, one-click source access and the security and compliance page are all what you build for users who are accountable rather than curious.

The named institutions also make the claims checkable in principle, which is a higher standard than a logo wall. What the site does not give is measured outcomes — how much time was saved, or how the tool changed what those researchers produced — so the case studies establish who uses it rather than what it achieved.

Researching Something You Have To Defend

Research tooling fails when it becomes a substitute for reading rather than a way of finding what to read.

This sequence uses the citation architecture for what it is actually good at.

1. Pick the right agent — search for facts, Scholar for literature, write for drafting.
2. Open at least three citations on any claim that matters.
3. Use the mind map first when a topic has many moving parts.
4. Treat a literature review as a reading list you still have to read.
5. Verify claims inside the draft rather than promising to check later.
6. Keep the sources with the document, not just the conclusion.