How the landscape feeds your brief

The literature landscape is not a separate report you open on the side. When the agent maps the biology your question touches, it turns that map into narrative and weaves it, citation-verified, directly into your brief.

#From map to narrative

Before it writes, the agent builds a map of the field around your question. Each paper it has read is reduced to the biology it actually discusses — genes, proteins, drugs, diseases, pathways, and controlled-vocabulary terms — and papers whose biological fingerprints overlap are connected. That map is genuinely useful on its own: community detection surfaces the research themes a field clusters into, anchor papers and the bridges between themes stand out, and each theme is marked as rising or fading.

A map, though, is not an answer. Structure tells you where the field concentrates; it does not tell you what the field has learned or where your question sits inside it. The synthesis step closes that gap. The agent reads the topology — clusters, hubs, bridges, thin regions — back against the full text it has already gathered and produces a connected account of the field: what the major lines of work are, which papers hold them together, where distinct areas meet, and where the evidence runs out.

One brief, not two artifacts

Earlier versions produced the synthesis as a stand-alone “landscape story” you generated with a separate action. That is gone. There is no separate button and no separate document — the landscape now lives inside the brief the agent returns, as prose and citations alongside the rest of the answer.

#What the synthesis covers

The synthesis is drawn from the same evidence base the agent used to answer your question, so it speaks to your field rather than a generic overview. Depending on what the literature supports, it can address:

Field overview. A high-level read of the research area your question touches — its scope, its current state, and how your question sits within it.
Major themes. The research clusters the map revealed, described in plain terms: what each line of work is about and which papers define it.
Anchor papers. The foundational, highly connected papers a theme is built on, and why they matter to the field.
Bridges. Cross-domain connections where otherwise separate themes meet — frequently where translational or unexpected opportunities appear.
Temporal trends. How the field has moved over time, including which themes carry momentum and which are emerging.
Evidence quality. How robust the supporting work is — whether findings replicate and how strongly the field agrees.
Gaps. Thinly evidenced or underexplored areas that the structure of the map makes visible.

The agent does not force every heading into every brief. If the evidence for a temporal trend or a bridge is not there, it says so or leaves it out rather than manufacturing a claim to fill the outline.

No lens selector

The synthesis does not ask you to pick a “fundamental,” “applied,” or “convergent” lens. There is no mode switch. The agent works from your question and the evidence it gathered — mechanistic detail, translational relevance, and cross-domain integration all surface when the literature supports them.

#Full-text grounding and citation verification

The synthesis is built from curated, full-text sources drawn from the life-science literature (40M+ papers), not from abstracts skimmed at a distance. That constraint is deliberate: only papers the agent could read in full contribute to the narrative. A paper that appears on the map but whose full text was unavailable does not get to shape claims in the synthesis, so the account is grounded in complete methods, results, and discussion rather than a title and a one-paragraph summary.

Grounding does not stop at “full text was available.” Every claim the synthesis makes is tied to specific papers, and each of those citations is verified against the passage it points to — the same full-text citation verification applied to the rest of your brief. A statement about a theme or a bridge carries the sources that support it, and those sources have been checked to actually say what the claim attributes to them. If a citation cannot be matched to its source passage, it is not presented as full-text-verified.

Richer input, richer synthesis

The more of your evidence base is available as full text, the more the agent has to synthesize from. Fields and questions where full text is widely accessible produce a denser, better-connected narrative; where full text is sparse, the synthesis is correspondingly more cautious and names the limitation.

#How themes, anchors, bridges, and gaps enter the brief

Because the synthesis is woven in rather than bolted on, its findings show up where they are relevant to your question rather than as a detached appendix. In practice:

Themes give the brief its structure — the agent organizes its account around the real clusters in the field instead of an arbitrary outline, so related evidence sits together.
Anchor papers are called out as the load-bearing sources they are, which helps you see quickly what a conclusion ultimately rests on.
Bridges point to where your question connects to adjacent work — often the most useful part of a brief, because it is where a finding in one area becomes leverage in another.
Gaps are stated plainly as gaps: thin or missing evidence, unresolved contradictions, and underexplored areas. These are the openings the agent draws on when it proposes hypotheses and experiments.

The result is a brief that reads as one connected answer: the landscape gives it shape and context, the evidence typing and citation verification keep every claim accountable, and the gaps it exposes feed directly into the next steps the agent suggests.

A synthesis, not the last word

The landscape reflects the papers the agent read for your question, weighted toward what was available in full text. It is a faithful account of that evidence base, not a complete census of the field. Treat rising and fading signals, bridges, and gaps as well-grounded orientation to pursue — not as settled fact independent of the sources cited alongside them.

← PreviousTrends & momentum Next →Evidence typing & weighing