§0 · LivingMeta

Tens of thousands of papers.
One researcher. Twelve months.

Systematic reviews take twelve to eighteen months — and the literature has moved on before they publish. Screening drowns PhD students. Extraction stalls. Gap identification stays subjective. Theses, dissertations, and grant proposals queue up.

AI is the obvious help — and the obvious risk. LivingMeta is built around a single principle: the question is no longer whether AI belongs in research, but how to use it so the work stays clean, controllable, and traceable. Six AI perspectives extract the same paper independently, you see where they agree and where they disagree, every claim cites the source verbatim, and the human reviewer has the final word.

See the demo →Visit a live instance ↓Request your field →

§1 · The work that doesn’t scale

Six familiar bottlenecks

The motivating pains behind LivingMeta. If three or more of these sound like your week, the rest of this page is for you.

P01

Stale reviews

By the time a systematic review publishes, eighteen months of new literature is missing. Update workflows are painful enough that most reviews are simply allowed to age.

P02

Screening that drowns students

An interdisciplinary query returns 50,000 to 200,000 hits. Title-abstract screening alone burns months of doctoral work.

P03

Extraction that doesn't scale

Reading 200 papers by hand to pull out methodology, effect sizes, sample design, and statistical tests is a full-time job behind every meta-analysis. Skip a paper, miss a moderator. Cut corners, inherit the bias.

P04

Theses, proposals, reviews — all laborious

Half a dozen theses at once, each one guided from “where do I start?” to a defensible gap claim. Plus grant proposals, peer reviews, lit-review chapters — the same scaffolding work, week after week. AI could plausibly do parts of it, and you don't dare.

P05

AI tools that don't add up

Elicit for extraction. Consensus for search. Cochrane for SR. A chatbot for the rest. None of them share an audit trail. None of them are reproducible end-to-end.

P06

AI that invents citations

Generic chatbots hallucinate DOIs and authors. The cost of a single fabricated reference in a published paper makes most researchers refuse AI for literature work — exactly when they need it most.

§2 · What LivingMeta does about it

One pipeline, four working surfaces

Papers

Continuously screened, transparently classified

Every paper in the field, scored for relevance, with the classifier decisions visible and overridable. The screening backlog goes away; what remains is judgement work.

Resources

Datasets and instruments, surfaced and FAIR-scored

Datasets, validated questionnaires, code repositories, APIs and measurement instruments extracted from across the corpus, each linked back to the papers that use them.

Agenda

A Priority Research Agenda derived from the literature

The “further research” sentences in tens of thousands of papers, aggregated, clustered, and scored for impact and feasibility. A defensible answer to “where should we be working next?” — updated as the field updates.

The Lab

Where you and an AI agent work together

Threads scaffold the eight phases of a research project, from exploration to writing. Role-adaptive coaching for laypeople, juniors, seniors, and supervisors. Every claim the agent makes is grounded in a paper in the corpus — by identifier, with the verbatim quote.

Read how it works in detail →

§3 · The professional middle

From “should I use AI?” to “how do I use AI responsibly?”

“The question is no longer whether AI belongs in research,” Janssen, Van der Kleij, and De Keijzer wrote in Sociale Vraagstukken earlier this year, “but how to stay clean, controllable, and traceable when you use it.”

They name two ineffective reflexes: paralysis (“don’t touch it, something will go wrong”) and overestimation (“AI can do everything, outsource your thinking”). The professional middle is craftsmanship — the human stays responsible, technology is deployed deliberately and within bounds.

LivingMeta is what that middle looks like in software.

Janssen, J., Van der Kleij, R., & De Keijzer, A. (2026). Wetenschap moet AI leren en durven gebruiken. Sociale Vraagstukken.

Decisions stay human.

The agent surfaces evidence; you write the analysis, the conclusions, the manuscript.

AI does what AI is good at.

Ordering, structuring, retrieving passages — never deciding.

Every step is traceable.

Every extraction carries the source quote. Every agent answer cites the paper by identifier. Every correction is logged.

The data is private and yours.

No consumer chatbot, no leaked corpus. Per-instance isolation, proper access controls, downloadable static JSON if you ever want to fork.

§4 · Living agenda · right now

A research priority, generated from the literature this week

Each instance produces a Priority Research Agenda from the “further research” sentences in its own corpus. Here is one item drawn from a current instance — representative of what every instance ships with on day one.

From menstrual-health-work.livingmeta.ai

Priority #2 · 87 papers

Longitudinal evidence on workplace policy interventions

The literature converges on a clear set of workplace interventions — flexible leave, on-site facilities, manager training, written policy — with 87 papers reporting on at least one of them. Only six follow outcomes beyond twelve months. The intervention base is broad; the durability evidence is missing.

Gap type
Methodological · Empirical
Suggested design
Longitudinal cohort, 24+ months, mixed methods
Matched resources
3 validated absenteeism scales · 2 disclosure instruments
Feasibility
Medium · funding-dependent
Open this gap in the Lab →See the full agenda

Every featured field has its own agenda — browse them via the field cards below.

§5 · Where LivingMeta is running

Six fields. Real corpora. Named anchors.

Every featured field is owned by a working researcher and built from an open academic corpus. Visit to browse, request access to use the Lab.

Sports Analytics

~120K papers

RSM Erasmus University Rotterdam

Performance analysis, tactics, player evaluation, and betting markets — across every sport and methodology.

Visit site →Request demo access

Innovation Management

~170K papers

TU Delft — Technology, Policy and Management

Innovation, entrepreneurship, technology commercialization, and the diffusion of new high-tech products.

Visit site →Request demo access

Management Consulting

~110K papers

VU Amsterdam — M&O

Consulting engagements, professional service firms, and the evidence behind consulting interventions.

Visit site →Request demo access

Public Governance

~70K papers

VU Amsterdam — FSW

Governance, integrity, public values, and accountability in the public sector.

Visit site →Request demo access

Emergent Organizational Change

~55K papers

Open University of the Netherlands

Sensemaking, power dynamics, and organizational development in complex public and professional organizations.

Visit site →Request demo access

Menstrual Health at Work

~130K papers

Period Media

The intersection of menstrual health, menstrual disorders, and menopause with the workplace — absenteeism, policy, disclosure, stigma, and the legislation shaping cycle-aware employers.

Visit site →Request demo access

§A · Foundation paper

From Search to Synthesis: A Living Research Platform for AI-Augmented Evidence Mapping

VU University Amsterdam / Erasmus University Rotterdam

The full argument — methodology, validation, architecture decisions — lives in the paper. Reach out and we send it over.

Request PDF — info@livingmeta.ai

§6 · Your field

It works for sports analytics. It works for menstrual health at work. It can work for your field.

A dedicated subdomain. We build the corpus, run the pipeline, and hand you the keys. The data stays yours; the workflow stays reproducible. We use existing open academic sources (OpenAlex, arXiv, thesis repositories); if you have a domain corpus of your own, we can plug it in alongside the literature.

Or email us directly at info@livingmeta.ai.