The living evidence brain for soil health, nutrient density and human health.

One evidence chain. Six connected fields. Read end to end.

Practice affects soil. Soil affects production. Production affects food composition and nutrient density. Nutrient density affects diet. Diet affects health. Pulse Brain treats every record as part of that chain — not as an isolated paper in an isolated discipline.

A record about mycorrhizal colonisation, a meta-analysis of organic vs conventional dairy, and a longitudinal study of dietary fibre and bowel disease all live in the same governed catalogue — and Pulse weighs them together.

A disciplined intake process — not an open web crawl.

Pulse Brain is built around a citation acquisition and maintenance protocol that runs every hour and is audited end-to-end. New evidence enters through multiple routes — and every candidate record is assessed against Vitagri's subject scope before it joins the live resource.

1. Discovery. Crossref & DOI search, PubMed and biomedical sources, OpenAlex metadata and reference graphs, Semantic Scholar, trusted publisher feeds, industry and policy research, manual bulk upload, citation snowballing, and coverage-gap audits.
2. Claims-check feed-in. Useful records discovered during a Pulse Claims Check review enter the same intake queue — so the resource learns from every uploaded document it grades.
3. Correction & deduplication. Candidate metadata is reconciled across sources, duplicates are merged, and obvious errors (truncated PMIDs, dropped DOIs, wrong year) are flagged and corrected.
4. Subject-scope assessment. Each candidate is judged against the soil-to-health scope. Out-of-scope records are archived in a rejection store with an audit trail — not silently dropped.
5. Tiering. Records are graded against an evidence-tier model: systematic reviews and meta-analyses at the top, through RCTs, cohort studies, observational studies, field trials, narrative reviews, policy reports and industry sources.
6. Enrichment. Approved records move through enrichment layers — claims, summaries, effect sizes, study design, geography, funder/COI extraction, open-access status, retraction status, citation velocity, reference graph and cross-record corroboration.
7. Semantic indexing. Records are embedded into a vector index so Pulse can retrieve by meaning — not just by keyword.
8. Projection & serving. The durable citation record stays canonical. Search indexes, retrieval views and semantic vectors are maintained as projections that can be rebuilt under controlled procedures.

Find out how Vitagri's Pulse resource is far more than a title and a reference.

Once a record is approved, it moves through enrichment layers that turn a citation into something Pulse AI can actually reason over — and that a farmer, a buyer, a funder or a policy team can actually use.

• Plain-language summary — a 2-3 sentence non-technical synthesis any reader can absorb.
• Extracted claims — 3-6 specific, testable claims pulled from the abstract, ready to be checked by Pulse Claims Check.
• Effect sizes — sample size, primary effect, direction, confidence interval, p-value, extraction confidence.
• Topic, theme & study design — taxonomic tags so the record can be filtered, mapped and visualised.
• UK applicability & key measures — practical relevance for the British farming and food system.
• Funders & conflict-of-interest signals — surfaced from the abstract as neutral metadata, not badges.
• Open-access status, licence & retraction state — refreshed on a scheduled cadence so the record never goes stale silently.
• Cross-record corroboration — pair-level verdicts on which other records support or contradict this one's findings.
• Citation velocity & reference graph — global citation count, two-year velocity and the catalogue's own forward & reverse reference links.

This is what allows Pulse AI to answer with evidence context rather than just a list of papers — and what allows Pulse Lens to show where evidence is dense, where it is thin, where the literature agrees, and where commissioning new work would be most valuable.

Six things — driven by one governed catalogue.

Pulse Brain is the evidence layer. Everything below it is built on the same trusted catalogue — so a claim graded by Pulse Claims Check, a map rendered by Pulse Lens, and an answer returned by Pulse AI are all reading from the same governed source of truth.

Powerful because it is honest about uncertainty.

Pulse Brain shows evidence tiers, uncertainty, gaps and contested areas. We do not say 'AI knows everything', 'instant truth', 'proves the link' or 'settles the science'. We say maps the evidence, weights the evidence, shows what is known and unknown, and connects research to action. Pulse's credibility comes from disciplined caution as much as scale.

That is also why Pulse is built on a transparent protocol — durable citation records as the source of truth, projections that can be rebuilt under controlled procedures, scheduled health checks that audit the system's own state, and a learning agent that surfaces failure modes before they reach a member. The full Citation Database Protocol governs every step. The page you are reading is the public-facing summary; the protocol itself sits behind the resource for those who need to audit it.

Pulse, nutrient density & soil-to-health — answered.

The questions farmers, food businesses, funders, researchers and policy teams ask most often when they first meet Pulse. Each answer is grounded in the same governed evidence catalogue that powers the rest of the resource.