Research Field Atlas
Where your next grant actually moves a research field — the under-served edges where money has leverage, versus the crowded core where it doesn’t — and where the field leans dangerously on a few institutions.
A research field is never a flat map. Some corners are crowded and well-funded; others are wide open and under-served. This report shows you where your money actually moves the field. It marks the saturated core where new grants return little, the under-attended edges where they go further, and the places where the field leans on a handful of institutions — a dependency worth knowing before you fund. And it holds a hard line on honesty: it tells you what the evidence shows, names what it doesn't, and offers the deeper reads as next steps you can commission — never as guesses dressed up as findings.
We tell you what the evidence shows — and, just as clearly, what it doesn't. Where a deeper read is needed, we name it as a next step you can commission, not a guess we slipped in to sound smarter. You'll never pay us for a confident-sounding claim the data can't carry. That line is the whole point: it's what makes every claim we do make one you can act on.
Built from publication, citation, and patent data across more than two hundred million scholarly works (OpenAlex, Semantic Scholar, plus the relevant publisher-side bibliographic stacks). Quote-only; typical commissions run $3,750–12,500. Built for foundation program officers, government science agencies, corporate R&D strategy, vice provosts for research, and policy think tanks allocating science funding.
Where does a grant actually move this research field — and where does the field lean dangerously on a few institutions?
What buyers actually ask.
Where does a grant actually move this research field?
The Atlas marks two structural facts: how saturated a part of the field is (a crowded, well-resourced centre returns little on new money) and how concentrated it is (a narrow institutional core is a structural dependency). It ranks the field's concept clusters, citation magnitude, and institutional output, then turns that into where new funding or attention has the most leverage — not just whether the field is active.
Does the Atlas map who cites whom, or name the bridge researchers who connect the field?
Not in the base deliverable — and we say so plainly rather than implying otherwise. The Atlas maps where the concept clusters, the most-cited work, and the institutions sit, and measures how heavily each work is cited, fully. The next layer — which work cites which, and the bridge researchers who connect different parts of the field — is a heavier pass we name as a commissionable next step, not something we quietly skip.
How big is the corpus you build the atlas from?
OpenAlex alone covers more than two hundred million scholarly works. We pull the field-relevant subset and map the structure across that corpus; pulls are capped, so every count is reported as a floor, not a complete census.
How is institutional concentration measured?
It's an output-and-citation concentration read: which institutions produce the field's work and its most-cited output, aggregated from the scholarly-publication record. It is not a funder/grant-metadata read — that data is frequently unusable across corpora — so the Atlas does not claim a funding-flow map unless the field's corpus supports it.
How is this priced?
Quote-only. Typical commissions run between $3,750 and $12,500 depending on field breadth, time window, and the depth of policy-grade methodology appendix the buyer needs. Scope is set during intake.
Why quote-only and not fixed-fee?
Field breadth varies materially. A niche area with five hundred active researchers and a broad cross-disciplinary field with fifty thousand active researchers carry different data and analyst budgets. We scope each commission against the actual corpus rather than charging the same for both.
How is this different from a Web of Science or Scopus report?
Bibliometric platforms produce an analyst's database; the Atlas produces an artifact. Citation-magnitude ranking, institutional concentration, time-evolution structure, gold-standard patent-IP translation, and the policy-grade appendix are the analytical layer the platforms do not produce.
How long does the engagement take?
Typically four to eight weeks end-to-end. Corpus pull and clustering are the long-pole steps; methodology appendix drafting and review extend the back end.
The deliverable, in detail.
- Concept-cluster structure: how the field's corpus organizes into a dominant core and its specialist tail, so you know where the crowded centre sits and where the under-attended edges are.
- Citation-magnitude ranking: the field's most-cited, foundational works — its gravity wells — read as a structural signal, not a verdict on scientific correctness.
- Institutional concentration: the compact set of institutions producing the field's most-cited output, with citation-footprint outliers named — the map for scouting talent or partnership.
- Time evolution: the field's publication trajectory since its founding era, cross-checked against an independent current-activity pull so "still active" is corroborated, not assumed.
- Patent-IP translation read at gold standard — family-deduplicated, holder-harmonized — showing who has actually translated the field into protected IP, a different map from who publishes it.
- Policy-grade methodology appendix suitable for external citation, documenting every source and every analytical choice.
How the report is built.
The Research Field Atlas begins with a research field, time window, and institutional or geographic scope at intake. The corpus is pulled from OpenAlex (>200M works), Semantic Scholar, and the publisher-side bibliographic stacks relevant to the field; deduplication and concept-cluster tagging follow.
Concept clustering runs on title-and-abstract semantics joined to citation co-citation structure, organizing the corpus into a core cluster and a specialist tail rather than a flat keyword list. Citation-magnitude ranking identifies the field's most-cited works — its gravity wells — read as where attention concentrates, not as a statement about scientific correctness.
Institutional concentration aggregates the field corpus by producing institution, with each institution's cumulative citation footprint layered in — surfacing the compact leadership set that produces the field's most-cited output, and naming where that output leans disproportionately on one institution's citation footprint relative to its volume.
Time evolution is read from the corpus by publication year, cross-checked against an independent current-activity confirmation pull so the read on whether the field is still active — not just historically large — is corroborated rather than assumed. The patent-IP-translation leg reads the patent record at full strength: global, family-deduplicated so one invention counts once worldwide, classification-and-full-text scoped, with assignee names harmonized — exact distinct-family counts, not a floor.
A senior analyst — typically with field-specific training — reviews each layer before the atlas is drafted. The Counter-Signal Pass on a research field surfaces methodological pushback, replication-crisis flags, and citation-cartel risk for any named cluster the atlas centers.
The policy-grade methodology appendix documents every source, every join, and every analytical choice so the deliverable can be cited externally.
Counter-Signal Pass is included on every report. The full Foragentis methodology is documented in The State of AEO and GEO in 2026.
What this report does NOT do.
Procurement-grade reports scope themselves. The work below is adjacent and important — and is not in this SKU.
We map the structure and the pace. The judgment call — will it work, will it win — stays with you and your experts. That's deliberate: it's exactly what makes every other read in the report one you can trust.
The who-cites-whom map — which work cites which, and the bridge researchers who connect different parts of the field — is a deferred next step, not a data gap. It needs more computing than a base engagement scopes; we name no bridge researchers and draw no connections because that layer was not built. It is available as a follow-on commission.
The base engagement does not include funder/grant-metadata parsing — that data is frequently unusable across corpora — so the institutional read is an output-and-citation-concentration read, not a funding-flow read. A funder-metadata-bearing pull is available as an addition where the field's corpus supports it.
Author-affiliation and patent-assignee data is messy in any open corpus. Institution and holder-name harmonization runs at industry-best precision but is not perfect — a few large filers' sub-entities are consolidated by hand.
Field-of-study tagging is a model-driven step. For interdisciplinary fields where the boundaries are themselves contested, the atlas reports the boundary choice it inherited and offers a sensitivity analysis on the boundary case.
What the engagement costs.
The Counter-Signal Pass — every thesis stress-tested against its strongest opposing case — is included on every report at no extra cost. See the Counter-Signal block on the catalog hub →
Adjacent reports.
Frontier Concept Brief
Academic concepts moving fast in the literature with no productized offering yet.
R&D Velocity Audit
Does the innovation actually ship? For a company: how fast it turns patents into shipped product, versus the peers you name. For a research field: which parts of the race are pulling ahead, and which are hype. Pick your read at checkout.
About Foragentis.
Foragentis is an AI research and product company based in Sacramento, California. ForIntel is the business-intelligence research arm — producing custom dossiers across four buyer lanes: Search & AI Visibility, Markets & Locations, Capital & Innovation, and Specialty.
Every claim in a ForIntel report traces to a public source. Findings are re-verified before delivery. The Adversary/Analyst architecture pairs a senior analyst with a counter-signal pass on every thesis. Anything below our statistical thresholds is labeled directional rather than validated.
Methodology is documented in The State of AEO and GEO in 2026 — a 9,900-word, 42-page public study with effect-size statistics across four frontier AI engines.
Ready to scope this commission?
Intake takes under five minutes. We confirm scope, timeline, and cost within one business day.